{
  "schema_version": 2,
  "period": "28 September to 6 October 2026, Asia/Singapore",
  "machine": {
    "model": "Mac Studio M3 Ultra",
    "gpu_cores": 80,
    "memory_gib": 512,
    "os": "macOS 26.3 (25D125)"
  },
  "scope": "Selected measurement fields from the session reports. Full traces remain local. This file is not a complete raw-log archive.",
  "limits": [
    "Qualification uses each source campaign protocol. The protocols differ.",
    "Decode rates exclude prefill. The time to the first token includes prefill and any measured pause.",
    "An incomplete request has no final decode rate. A missing value is null.",
    "Retrieval tests do not establish general model quality. Manual review found additional defects.",
    "The early TensorFold Nemotron, Flash-Next, and GLM tests use a temperature-1 sampler.",
    "The final Qwen3.8 and TensorFold 0.6.5 GLM tests use a greedy sampler.",
    "The dataset includes all attempts in the three combined summaries, plus the listed earlier measurement summaries.",
    "The export includes the new GLM startup failures. Complete setup logs and invalid mlx.fast arms remain local.",
    "The earlier E9 interruption has no numeric export row."
  ],
  "best_qwen_timing": {
    "source": "qwen38_variants_bench/results/run_20261004_225650/tf_flash_260864_1234_tensorfold_mtp_1/result.json",
    "first_token_seconds": 735.4590765829198,
    "decode_seconds": 10.82510908297263,
    "measurement_wall_seconds": 746.3316863330547
  },
  "sources": [
    {
      "path": "qwen38_variants_bench/all-results.json",
      "sha256": "f1031b5d586ad63bcd6b96933a3ac6aba3cec68e742d431a1ec9fcd34882cc12"
    },
    {
      "path": "qwen38_variants_bench/results/run_20261004_185731/27b_bare_2048_1234_mtplx_mtp_1/acceleration-comparison.json",
      "sha256": "0aa30b2685c7889f156b20bbc00cc46573101752c29149c420c482687930203d"
    },
    {
      "path": "qwen38_variants_bench/results/run_20261004_185731/27b_bare_32768_1234_mtplx_mtp_3/acceleration-comparison.json",
      "sha256": "317de48ffdde1ee0e952326aadd4f28e68405ad55daf206a1f968c85393f16a4"
    },
    {
      "path": "qwen38_variants_bench/results/run_20261004_202905/27b_speed_2048_1234_mtplx_mtp_1/acceleration-comparison.json",
      "sha256": "b6884a6edae7b4b2445895098fbae608b79d58bc61c78bc9170e725833379641"
    },
    {
      "path": "qwen38_variants_bench/results/run_20261004_202905/27b_speed_32768_1234_mtplx_mtp_3/acceleration-comparison.json",
      "sha256": "557a2ec2081d90c0af59d404ea8cda859e6b229c8bd82b17cc9d6015292f17c7"
    },
    {
      "path": "qwen38_variants_bench/results/run_20261004_211626/flash_bare_2048_1234_mtplx_mtp_1/acceleration-comparison.json",
      "sha256": "d4d92474849a336e08fe60732be8930da5217bf54b5069dea48164f49795d73e"
    },
    {
      "path": "qwen38_variants_bench/results/run_20261004_211626/flash_bare_32768_1234_mtplx_mtp_3/acceleration-comparison.json",
      "sha256": "7e6602e28c72b80e0421f9f7c2eda6fbf77953a6d48ba4b004ed693bb7dd1e69"
    },
    {
      "path": "qwen38_variants_bench/results/run_20261005_001052/flash_speed_2048_1234_mtplx_mtp_1/acceleration-comparison.json",
      "sha256": "c2ec3766d7df75fcc36439b542d73ee27f4a15711bd3aab6dba0d1f864515ef0"
    },
    {
      "path": "qwen38_variants_bench/results/run_20261005_001052/flash_speed_32768_1234_mtplx_mtp_3/acceleration-comparison.json",
      "sha256": "0b4f965b5a6d978e8c9702b997c497c380509aad3dfbc9a0073e8b2f5f021ec2"
    },
    {
      "path": "qwen38_variants_bench/results/run_20261005_014554/flash_speed_260864_1234_mtplx_mtp_1/acceleration-comparison.json",
      "sha256": "8795d9467c9116f4bb47076a62ca6b6100f2dab73cb69d12c878fa6149de1f65"
    },
    {
      "path": "qwen38_variants_bench/results/run_20261004_215711/tf_flash_2048_1234_tensorfold_mtp_1/acceleration-comparison.json",
      "sha256": "e0bcd775b4e53c7b6f465bb455f9f68ac4a5822b84d1eb795c1feb183283d08d"
    },
    {
      "path": "qwen38_variants_bench/results/run_20261004_233659/tf_flash_32768_1234_tensorfold_mtp_1/acceleration-comparison.json",
      "sha256": "690e937f8c4459a1877e655c81baabfd94990685a8162e853e15847d6571ff7b"
    },
    {
      "path": "qwen38_variants_bench/results/run_20261004_215711/tf_flash_32768_1234_tensorfold_mtp_3/acceleration-comparison.json",
      "sha256": "a8e57ccef798ce2ada7009296037b2fa42e9dcf9660c9666272e274f6a8d741e"
    },
    {
      "path": "qwen38_variants_bench/results/run_20261004_225650/tf_flash_260864_1234_tensorfold_mtp_1/acceleration-comparison.json",
      "sha256": "8f71a02fa82af8c75d2d1c5afd53ad1e15415bd3505210470b04a3621904d1e2"
    },
    {
      "path": "qwen38_variants_bench/results/run_20261004_215711/tf_flash_260864_1234_tensorfold_mtp_5/acceleration-comparison.json",
      "sha256": "76be096ebde6cdd84a32a395fc349ce323b7b2b0311cd25316259184309d0e88"
    },
    {
      "path": "eight_models_bench/all-results.json",
      "sha256": "02038f190ad95f9451615cf314f117b5031561cccac8834e1a227f3fee8f99e7"
    },
    {
      "path": "eight_models_bench/results/run_20261002_173233/gemma26_2048_1234_gemma-vlm_mtp_8/acceleration-comparison.json",
      "sha256": "f839b259e240e58f924eb25fd7bca39f61467e9690e8e63ea15d724dd9c60a63"
    },
    {
      "path": "eight_models_bench/results/run_20261002_191509/gemma26_2048_1234_gemma-vlm_mtp_1/acceleration-comparison.json",
      "sha256": "78c0db7ba7ca865e67c03e72e2316fe1743c157756c3d8144051bd5cba084d79"
    },
    {
      "path": "eight_models_bench/results/run_20261002_191509/mimo9_mtplx_2048_1234_mtplx_mtp_3/acceleration-comparison.json",
      "sha256": "f9d5be7d039a850caf2acf6717fb34bb72da1aab0e60e727de60d2f29fb748d7"
    },
    {
      "path": "glm53_tensorfold065_bench/all-results.json",
      "sha256": "60a5999b3f108f81b6941771f3c23bf136b3fffdc518e423ece44059c16ee1e8"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_142827/glm_4bit_2048_1234_tensorfold_serial_0/result.json",
      "sha256": "a7fa7770b630d274662f195b4dae44883c1ea7fb1175ef66d6385a91d4a20f1e"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_145619/glm_4bit_2048_1234_tensorfold_mtp_1/acceleration-comparison.json",
      "sha256": "0a172ab62fc1023c1e23ced602b38a1c8b4b90265f378a7b84a3103354ab17e7"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_145619/glm_4bit_2048_1234_tensorfold_mtp_1/result.json",
      "sha256": "84a54e69155f7447d515751e1337647c9720d9688d7283cd10e9a8e9a8f2725d"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_145619/glm_4bit_2048_1234_tensorfold_serial_0/result.json",
      "sha256": "5c2783800801e78a4e7a3ebf2f87c845630c83ed0feb427f950d1923fc9ba23a"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_145619/glm_4bit_260864_1234_tensorfold_mtp_5/acceleration-comparison.json",
      "sha256": "47c20eb7ca78bbf05ae89ae5e7d914d208efdf8b047df61d6908d86415a54205"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_145619/glm_4bit_260864_1234_tensorfold_mtp_5/result.json",
      "sha256": "81366d70efe698e797cd48b2a632c4df2b7cc891769bb76981b14790af9322ad"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_145619/glm_4bit_260864_1234_tensorfold_mtp_7/acceleration-comparison.json",
      "sha256": "0a0671235a963221e58c57e853dd39c80701df15720d7f1aec65666bbbb4be35"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_145619/glm_4bit_260864_1234_tensorfold_mtp_7/result.json",
      "sha256": "fcb483f8c655a0b37a47b8384c0c4bac847767ef9d6c8cec2dbeba828846ace7"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_145619/glm_4bit_260864_1234_tensorfold_serial_4/result.json",
      "sha256": "e6a90599b29828a72275a773a68036ff1d70bb4b512ee393f3b09fda0c284151"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_145619/glm_4bit_260864_1234_tensorfold_serial_6/result.json",
      "sha256": "0bfda7c33b4c684ed69a3575690309f1b2b6d664769a067b6f2fd7be61f2bb40"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_145619/glm_4bit_32768_1234_tensorfold_mtp_3/acceleration-comparison.json",
      "sha256": "87ca05d7d3cd00341ebede309030d357faad3d672071aad0cf4fa36bd6f24d29"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_145619/glm_4bit_32768_1234_tensorfold_mtp_3/result.json",
      "sha256": "f92fabdb19a8067c2d68f711cd3aacbb81fd641fef6dbe29b03d438e88d65f1c"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_145619/glm_4bit_32768_1234_tensorfold_serial_2/result.json",
      "sha256": "ffcc48f132cd40c0967cae14a6769f8ff8672adcae6ed4d6a846bd5e2126a758"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_145619/glm_oq4_2048_1234_tensorfold_serial_8/result.json",
      "sha256": "0fcbaeb05cb0de551c6c7a2ec8f6af9621e85df12c980f9605d68454ecace5e8"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_145619/glm_oq4_2048_1234_tensorfold_serial_8/startup-failure-diagnosis.json",
      "sha256": "7f8529fbda7d5ba24f616d089610c8c9a2e4b91995e616e6819b1e87fda5ff67"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_200547/glm_oq4_2048_1234_tensorfold_mtp_1/result.json",
      "sha256": "ebc61ac1157bed146ecf2703bb92c512fe07c799904ff39fa7e04916394a4d78"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_200547/glm_oq4_2048_1234_tensorfold_mtp_1/setup-stop-decision.json",
      "sha256": "3c4fd790c1461c7d2fabf0b60a97e16d9160e0b1f9e9a60abfa4eb66e359e83c"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_200547/glm_oq4_2048_1234_tensorfold_serial_0/result.json",
      "sha256": "eaedf72b11f0eb2f80237df8b8e6361629f48fee478da5ce25eec11494d63f70"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_213646/glm_oq4_2048_1234_tensorfold_mtp_1/acceleration-comparison.json",
      "sha256": "8bf6186c36476d922bd2c683af3f15e26345a8b5ab5abf0b3bb9bb68825ca295"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_213646/glm_oq4_2048_1234_tensorfold_mtp_1/result.json",
      "sha256": "fdef58756be683a237c60f2016380c0a8b1affef1b750357b3b80357f4ce0343"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_213646/glm_oq4_2048_1234_tensorfold_serial_0/result.json",
      "sha256": "d0b41716038f71a70c68cee1472d7e55f2e177a3943ff2a38750c3d52b74ec23"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_213646/glm_oq4_2048_1234_tensorfold_serial_0/reference-comparison.json",
      "sha256": "78d223bd69cf12b77bfb4e21876938ee9ffed67f63372f2b193b35d66a402e9e"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_213646/glm_oq4_260864_1234_tensorfold_mtp_5/acceleration-unavailable.json",
      "sha256": "0a4a2dd32f55a8e1640923ebb756fd00d38b29e13ad9adafe12a08b6ec1bb69c"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_213646/glm_oq4_260864_1234_tensorfold_mtp_5/result.json",
      "sha256": "9dc3a666ca25c0b0cb7233f01a4cc0d2974eb0946139f3e0f84e18ede5bc03f1"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_213646/glm_oq4_260864_1234_tensorfold_serial_4/result.json",
      "sha256": "1d5c00ec405f19e962121ba489dd11fa7004a2158fa02e835e5ee28c55747087"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_213646/glm_oq4_260864_1234_tensorfold_serial_4/load-abort-observation.json",
      "sha256": "1c1447c85d9696720e8c7a8f49dcb9b22483cf3a2b65317f1fc5db96b62e90a5"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_213646/glm_oq4_260864_1234_tensorfold_serial_6/result.json",
      "sha256": "070c75f4cc42b841db13becfe85d4ceb3ce7e38eebac5c1b8d55460ab59a1662"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_213646/glm_oq4_260864_1234_tensorfold_serial_6/startup-failure-diagnosis.json",
      "sha256": "1b715adc92319cff4b295bc6560a5a5ac1da5ae5c6374d121e221ea61555df2a"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_213646/glm_oq4_32768_1234_tensorfold_mtp_3/acceleration-comparison.json",
      "sha256": "83599797cde3ba35faf2cb9c14da56c620c43379a7c90c54ae9a4cdeacd54f4b"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_213646/glm_oq4_32768_1234_tensorfold_mtp_3/result.json",
      "sha256": "6fd2b02cd94519a7e1bd92b808491019a0d4ccd0d32bc3878acae7f70a0b3c87"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_213646/glm_oq4_32768_1234_tensorfold_serial_2/result.json",
      "sha256": "b45a42c8c77edc37c20d1c3ef004030bd9ac86e43917c1d972ee17221b0c8908"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_230806/glm_oq4_260864_1234_tensorfold_serial_0/result.json",
      "sha256": "6bafca6b7a0f8f2e35fd7a8dd5e61616460b7dd6a8d9c9649324334d135a7978"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_230806/glm_oq4_260864_1234_tensorfold_serial_0/startup-failure-diagnosis.json",
      "sha256": "d23b1fbc10d75a29bbff4df17bb4fda09698540d926a5f17cc080919cabf48c6"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_232901/glm_oq2_2048_1234_tensorfold_serial_0/result.json",
      "sha256": "38b361be6cdb127fe0e84e187c35f934359f8259d292d6c6e56494d207dc680b"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_232901/glm_oq2_2048_1234_tensorfold_serial_0/startup-failure-diagnosis.json",
      "sha256": "f335b6b4e7697ed7a64ce249cf08d406e47c46160a085b04a6ed7c6cb9a35878"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_234048/glm_oq4_2048_1234_tensorfold_serial_0/result.json",
      "sha256": "16c7fa07a96add957a2c34bd1a1cd68b4425deecc1f13f0019de3fb3f4833dd1"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_234048/glm_oq4_2048_1234_tensorfold_serial_0/loader-variant.json",
      "sha256": "5cf96b7670909e3288e5abc14a3bd4294bf1ccb073c55473a4d4870e0f1b335d"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_234048/glm_oq4_2048_1234_tensorfold_serial_0/reference-comparison.json",
      "sha256": "5e10a57d789a1b0fee1ff45c5e78820c7c82fa2c71a160ca8f9f9910b3bc22cd"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_234048/glm_oq4_260864_1234_tensorfold_serial_2/result.json",
      "sha256": "4dec19a7227e4fc6989bd6ee9ca42a7814eab02bdae8a3d793c0baeaeedae156"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_234048/glm_oq4_260864_1234_tensorfold_serial_2/loader-variant.json",
      "sha256": "10a6ff27ef48a295912f6e25ed6df9effe8fafc2ef1402b484fab6c46a6fc430"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_234048/glm_oq4_260864_1234_tensorfold_serial_2/interruption-diagnosis.json",
      "sha256": "0460e40130de7d1efce221880718c4188fb9459862ad1051cd032b1851d291c2"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_234048/glm_oq4_32768_1234_tensorfold_serial_1/result.json",
      "sha256": "a57f01469d2de18c639604fd7584e0b001d005c11a6d4d80cccc1eae4ebb93a4"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_234048/glm_oq4_32768_1234_tensorfold_serial_1/loader-variant.json",
      "sha256": "192c15d63369ecc898e1db8a0540e10fdc101a395443284440aceeb9b117bb76"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261005_234048/glm_oq4_32768_1234_tensorfold_serial_1/reference-comparison.json",
      "sha256": "ca8d15986ccaf5c8c11e58e69094a6ce87d4dc11937c2aceb865e6ca02b9a1cb"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261006_025508/glm_oq2_2048_1234_tensorfold_mtp_3/acceleration-comparison.json",
      "sha256": "642dc5abb2db4c8284f610ff95e51ccf7e9e97f59d5fd613376e9cbb7d41241c"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261006_025508/glm_oq2_2048_1234_tensorfold_mtp_3/result.json",
      "sha256": "c3f79854cb666aa0915f4a13548faf8b2e21bd2d70bbbe058575eb4a763908f2"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261006_025508/glm_oq2_2048_1234_tensorfold_mtp_3/loader-variant.json",
      "sha256": "52ff813a33883ae59d2a2e0a3a8cb5e58ed171260167a6e560d3a0aa070418a9"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261006_025508/glm_oq2_2048_1234_tensorfold_serial_2/result.json",
      "sha256": "041ba974a28e8a4e79ec831ab68c144f841b2ba42053573ea8c9cd1f6777516c"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261006_025508/glm_oq2_2048_1234_tensorfold_serial_2/loader-variant.json",
      "sha256": "a3ace8d81245832da30e93db290a078d1a2fecee311cad4acb1b82e27a5a2fd6"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261006_025508/glm_oq2_260864_1234_tensorfold_mtp_7/result.json",
      "sha256": "fb9845a13ba5860daeb63b2696a3ff0d19de479d6395a8d9aecdb7941265cbc1"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261006_025508/glm_oq2_260864_1234_tensorfold_mtp_7/loader-variant.json",
      "sha256": "4813ab8546633bf51af943980c8340e4be7182f11662c3bb503b4717b3b64d8a"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261006_025508/glm_oq2_260864_1234_tensorfold_mtp_7/load-abort-observation.json",
      "sha256": "e2eb925b6f45af5b4c5e668fe31f3e6e94b8bebce572a5b990ec4022d9feb4c1"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261006_025508/glm_oq2_260864_1234_tensorfold_serial_6/result.json",
      "sha256": "78b17676ec82fb636c1f0e3d1c367667f19b37f2400318a9a4e823cb3c67550d"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261006_025508/glm_oq2_260864_1234_tensorfold_serial_6/loader-variant.json",
      "sha256": "e8cf8ef91dda2a2b7c1b8a72e8323a4ab9e9512335b3bd8f2b455c4c3d3f13a1"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261006_025508/glm_oq2_260864_1234_tensorfold_serial_6/load-abort-observation.json",
      "sha256": "823a980cd3a944da5d74b176e7b1c248d4e1b46f2958c26929f5558ba70fd040"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261006_025508/glm_oq2_32768_1234_tensorfold_mtp_5/acceleration-comparison.json",
      "sha256": "5de81ac69ae550434f28adc9a8984a7289154c2b563a64cafe0c5517d734ddf5"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261006_025508/glm_oq2_32768_1234_tensorfold_mtp_5/result.json",
      "sha256": "f6a8475d2c7b06d596751f5411759a60a548acef2b3eef16d9960192ce801b58"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261006_025508/glm_oq2_32768_1234_tensorfold_mtp_5/loader-variant.json",
      "sha256": "98e32afe4df4789b085caedadb8d34b5e97bc09a780957576d1bd1ca3a8646c8"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261006_025508/glm_oq2_32768_1234_tensorfold_serial_4/result.json",
      "sha256": "3728a0f231bee5c0a010361149d7769744a9f6aed3d9ed256b5f5215be91b18c"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261006_025508/glm_oq2_32768_1234_tensorfold_serial_4/loader-variant.json",
      "sha256": "1fcd4c20fa8ed306e8b4069a09567586a509c4eb0084d8308028b552bfeecd31"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261006_025508/glm_oq4_260864_1234_tensorfold_mtp_1/result.json",
      "sha256": "e6f6d820ca3d95b17dd7e8adf905e07cc1f86ff4876c67d5d92601d99e3916dc"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261006_025508/glm_oq4_260864_1234_tensorfold_mtp_1/loader-variant.json",
      "sha256": "773ccfbfed6bbd6d2b494d7cdbeac274968fdb0a6593e08aaf98128fec3d3975"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261006_025508/glm_oq4_260864_1234_tensorfold_mtp_1/load-abort-observation.json",
      "sha256": "edbbe17e094e0620fc451215ec793a0624919d38fe9273b9207b1c6b118e084f"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261006_025508/glm_oq4_260864_1234_tensorfold_serial_0/result.json",
      "sha256": "6a6ebd5daffc80c0a130ec6b60ceb11eef58afe1a2679becef37c6f756fc151a"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261006_025508/glm_oq4_260864_1234_tensorfold_serial_0/loader-variant.json",
      "sha256": "f4b5b83df472d02f7fe8daf7c17250cc5abd6432220acc36ec1587497d7cdcd0"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261006_025508/glm_oq4_260864_1234_tensorfold_serial_0/load-abort-observation.json",
      "sha256": "120cffe764e177899cd2d5828ff17fe87bef7d34be8e00638606a7edc31eca78"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261006_063141/glm_oq4_260864_1234_tensorfold_serial_0/result.json",
      "sha256": "9ebf08ccf803fc3c8902d697f12781827a1c4423dc809f325783d99beee26e26"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261006_063141/glm_oq4_260864_1234_tensorfold_serial_0/startup-failure-diagnosis.json",
      "sha256": "f4e3dbcde56bd6d4e9a7ac5e7515fdf4266f84d76b526146e0509c1037f02496"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261006_073714/glm_oq4_260864_1234_tensorfold_mtp_0/result.json",
      "sha256": "a40e84680716cdb14ea43f5f8f173fffef96dec5bc6364e57e9406dacc82e4bb"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261006_073714/glm_oq4_260864_1234_tensorfold_mtp_0/loader-variant.json",
      "sha256": "97824160510ac4c412af23b3fb427a87debb2a8573b28f822953bd241b9b04db"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261006_073714/glm_oq4_260864_1234_tensorfold_mtp_0/background-intervention.json",
      "sha256": "654a692ef6c48013efe24b0c5321398f304aaf872c7362dc2cb313b2ee9a1a83"
    },
    {
      "path": "glm53_tensorfold065_bench/results/run_20261006_073714/glm_oq4_260864_1234_tensorfold_mtp_0/load-abort-observation.json",
      "sha256": "05333f1d34152a9bba6285cfcf0cb41e6dd70bcaed1793856b27e9cdfcacc3a4"
    },
    {
      "path": "glm53_bench/summary.json",
      "sha256": "e6a048eb289df371880c68602fd228689a220436337e02eef88234e6764e1354"
    },
    {
      "path": "flashnext_bench/results/run_20260930_115151/summary.json",
      "sha256": "65c4eb45ed6950e9d19378c84cd7a4990b3f4d7c4441390ef3652cd293f24fc7"
    },
    {
      "path": "tensorfold_bench/results/run_20260930_000006/summary.json",
      "sha256": "27852559b94ddc0a2bbdfd63dd225eea11dac68e1f51d51f6bfbab5586bf5343"
    },
    {
      "path": "tensorfold_bench/results/run_20260930_000800/summary.json",
      "sha256": "e5ee62121bd02c0f51ec5fa5cdb30fa2e3dc66e656212045289cd67ed60c0365"
    },
    {
      "path": "tensorfold_bench/results/run_20260930_005309/summary.json",
      "sha256": "2d8835efce5046f8644a75ea10999583b284e816f5c215dbf334079e1538ca34"
    },
    {
      "path": "e9_bench/results/run_20260929_163156/summary.json",
      "sha256": "55f5fc2dee4cdcbeda111c4e3740b3cca404485c3735ede346be84c65cdbb1b8"
    },
    {
      "path": "qwen38_27b_bench/summary.json",
      "sha256": "d81fd44bf79dfdf169dbad3eca56a5d5cd5a0ce76de5b454e4fe8994c8c86791"
    },
    {
      "path": "qwen38_27b_bench/results/run_20261001_173331/summary.json",
      "sha256": "ef3639e05b0989aab98a307a745ccf556568ef42d04de972b3cbb6c2ff4ef29e"
    },
    {
      "path": "qwen38_27b_bench/results/run_20261001_180533/summary.json",
      "sha256": "28beed756cb21f8a184f94d58a1ba02e09b867579f0153d278fe8cebaa8fb581"
    },
    {
      "path": "qwen38_27b_bench/results/run_20261001_182755/summary.json",
      "sha256": "4412f43bdfa53ffd7397f784ae4db46329072588b20f96cddc7d43eba06c9f2b"
    },
    {
      "path": "qwen38_27b_bench/results/run_20261001_184030/summary.json",
      "sha256": "ca7cb69a7cc5b2e6e52dbef5234a191c6769f0423822309ff686795265db00fa"
    },
    {
      "path": "mlxfast_bench/results/c32k_fans/summary.json",
      "sha256": "9d5074cc540e274e4f4620699f2af4770f9badb2fba3bbd7a58dcd4af86649bd"
    },
    {
      "path": "qwen38_variants_bench/manifest.json",
      "sha256": "dbcd2dba86d7c8cbd8589d831ec7ceede3ffaa1323d4c6372cdd6cbed22ee613"
    },
    {
      "path": "eight_models_bench/manifest.json",
      "sha256": "e968e3ec418c7fdb5739a12a692bb8f2b3eab55445a0ba0839be9984a2adea83"
    },
    {
      "path": "glm53_tensorfold065_bench/manifest.json",
      "sha256": "2ca5581a2cd6ca601882955ddf5e68cb972e5682813f624f0b6c56582b9c92ea"
    },
    {
      "path": "qwen38_variants_bench/artifacts/tf_flash.json",
      "sha256": "5ecd4c19dfb749da2a6aafe3f38996b247cdc67ce1a6655f2e0bd667e5e2c1b4"
    },
    {
      "path": "glm53_tensorfold065_bench/weight-formats.json",
      "sha256": "1c5c0d615a4b5bf913596c494a9965f500ecd6b27492dab0836350a5646652f3"
    },
    {
      "path": "docs/28092026_mlxfast_32k_benchmarks.md",
      "sha256": "7204cab54712125b677ae8b5bcd1ea3b07ce315d5308fce2c833296d76195d99"
    },
    {
      "path": "docs/29092026_macmon_temperature_fix.md",
      "sha256": "d565919a794409c24081c339b9d8ce5958ec02777407a211e702ff33b1538419"
    },
    {
      "path": "e9_bench/results/run_20260929_163156/REPORT.md",
      "sha256": "02a0122e2d9244f9b1a72b0be84fb195249da98d4a49ccec4a42a1ad1c38ccf0"
    },
    {
      "path": "docs/20260929_tensorfold_nemotron_256k.md",
      "sha256": "cf69c54621becb9159c340bc4d9a4f510eb3d00113df90e1bf0da7fd3fd90fd1"
    },
    {
      "path": "docs/20260930_tensorfold_flashnext_256k.md",
      "sha256": "7ea752afd762f8c3dc59d9c74b669a1352908c3cad7270ef977c37bc0fe3860d"
    },
    {
      "path": "docs/20260930_glm53_flash_variants.md",
      "sha256": "021c5d91a83b16336459b8ba63363bb531424978c2b2de8508c480d9d068a88a"
    },
    {
      "path": "docs/01102026_qwen38_27b_mlx_benchmark.md",
      "sha256": "0d6a504c41d62d067c18c7697e53cb37bb1b5691d5c938c2822f2d14aee0843b"
    },
    {
      "path": "docs/02102026_eight_models_256k_benchmarks.md",
      "sha256": "4ef880abc60d77f1e73f52f9a19e06c07bf61e3273ad281cd2c831d89c10ee48"
    },
    {
      "path": "docs/04102026_qwen38_variants_256k.md",
      "sha256": "dee30bdd4fedce293a914939e1372197f3f63421585115d000ac1375e798553b"
    },
    {
      "path": "docs/02102026_256k_100tps_candidates.md",
      "sha256": "9be02253d1bf228e1a20d39b0a7bf003af9dd8a38d61e48b3ff95b24c29fdb7c"
    },
    {
      "path": "eight_models_bench/mimo/INVESTIGATION.md",
      "sha256": "461a66d9fb0688ef7726538da380d33939027da67976d2f5ee4351da5fea5308"
    },
    {
      "path": "qwen38_variants_bench/METHOD.md",
      "sha256": "0e016d5aeebdffbfe7324d4ef6b45092a3d456d3a724af121789f974b4569211"
    },
    {
      "path": "docs/05102026_glm53_tensorfold065.md",
      "sha256": "3055bbb956eda9d8e7c6b6619e21b48cd02e3de6379ff19423588aa88eb0da8b"
    },
    {
      "path": "glm53_tensorfold065_bench/README.md",
      "sha256": "08302ed80f9153548b30361f66efe993007e0688179d876e5191e482c09dfc45"
    },
    {
      "path": "glm53_tensorfold065_bench/tensorfold-source.json",
      "sha256": "d281e3547ce9e82703aa542a23544593d1bf65d81d475b8dea096a383f938965"
    },
    {
      "path": "glm53_tensorfold065_bench/protocol-comparison.json",
      "sha256": "d8608f11dfe98bff18c6b4ed9dd1b33e78cbb584d52341c760559624fd6dcdf2"
    },
    {
      "path": "glm53_tensorfold065_bench/direct-load-plan.json",
      "sha256": "8b557f20aabb4eb55809760cdc19b6bcd51fbbf6f5a31b39b0c80e5d9d09076c"
    },
    {
      "path": "glm53_tensorfold065_bench/direct-load-decision.json",
      "sha256": "3c4fd790c1461c7d2fabf0b60a97e16d9160e0b1f9e9a60abfa4eb66e359e83c"
    },
    {
      "path": "glm53_tensorfold065_bench/eager-load-plan.json",
      "sha256": "88b209891a569e2702e15925e013817af8147240a2562fba3102d8fc157d7939"
    },
    {
      "path": "glm53_tensorfold065_bench/eager-load-decision.json",
      "sha256": "7144b10a001c8fec722b3b5b8f17d3b913da8d93747d9c22b2540c68350dd09a"
    },
    {
      "path": "glm53_tensorfold065_bench/eager-loader-tests.json",
      "sha256": "6d7d6bfb07564816995e7845e66ea0587322e8f09139b1a78a75e9f7bf3d7e5d"
    },
    {
      "path": "glm53_tensorfold065_bench/eager_weights.py",
      "sha256": "5bcb0ec46a185d500aa65635dc034882059d37164b2b858b135c4f8e4e7186d6"
    },
    {
      "path": "glm53_tensorfold065_bench/loader-investigation/hypothesis.json",
      "sha256": "0dc9b400f687cde35e7dcce3aac021250901ac4e858f604b112a5039c366683e"
    },
    {
      "path": "glm53_tensorfold065_bench/loader-investigation/sources.json",
      "sha256": "5275649843f025f166cadd4641cf43b2cf32f08fb44a51295fb3bef3fd76ee9c"
    },
    {
      "path": "qwen38_variants_bench/tensorfold-source.json",
      "sha256": "d281e3547ce9e82703aa542a23544593d1bf65d81d475b8dea096a383f938965"
    },
    {
      "path": "qwen38_variants_bench/tensorfold-integrity.json",
      "sha256": "e48e2cb7bd40fba9769057d8e693824ff652f0cf696b7b28e000aa02b2971465"
    },
    {
      "path": "glm53_tensorfold065_bench/runtime-integrity-start.json",
      "sha256": "a67279ad02187e23e1b229d4296a094f8ea07ef077945f21c9e69e24a1569f91"
    },
    {
      "path": "qwen38_variants_bench/runtime-integrity.json",
      "sha256": "0b1e29d638f863686c6323a9c640341d03101ae4cabab747f98697dcccd74ca6"
    },
    {
      "path": "tensorfold_bench/requirements.lock.txt",
      "sha256": "60edae2c2d4c84b38508489563febbee318755373f89a0408c739d1473d93ec6"
    },
    {
      "path": "flashnext_bench/requirements.lock.txt",
      "sha256": "60edae2c2d4c84b38508489563febbee318755373f89a0408c739d1473d93ec6"
    },
    {
      "path": "glm53_bench/runtime-provenance.json",
      "sha256": "eddd34b198b4b2a13333314396d02f1f2a1dd5fa902ebabc2c97181ee43f50cf"
    },
    {
      "path": "glm53_bench/runtime.patch",
      "sha256": "13667562e648fae56db51a20b2633a7653cb5655bdc9709ce84c65a3079d3d66"
    },
    {
      "path": "eight_models_bench/runtime-integrity.json",
      "sha256": "23126377d8ab7cbe594c72792728ca104a17d9c936d6d0e93c40375f7f80a0fd"
    },
    {
      "path": "eight_models_bench/worker.py",
      "sha256": "90a1e215107bf104c6bb6aaa2cb70e15476fdf2d2922fe03afb1423f82351041"
    },
    {
      "path": "eight_models_bench/ling_adapter.py",
      "sha256": "2cab71f8ad08c24865c518389db9e395f35d41f83971737d4dad5ff6f94a86d6"
    },
    {
      "path": "eight_models_bench/ling/ADAPTER.md",
      "sha256": "2b13edfcee06367ae3e902c03d5441eb2607738e5dc3ed0f75e4968e158c2d79"
    },
    {
      "path": "eight_models_bench/accel_engines.py",
      "sha256": "97f0feece64535d1dc8e43a836523241b8d078d939810ace098e94401a4d584f"
    },
    {
      "path": "eight_models_bench/mtplx-runtime-integrity.json",
      "sha256": "5fbf3f8c0d850180e3c7fe52c00effa78b98ffc244962779890931fef12acd17"
    },
    {
      "path": "qwen38_27b_bench/runtime-version.json",
      "sha256": "da9b37189d08723c94d93841773f80d55ef31c5c49c2b2b60c10fdbc2df2dff4"
    },
    {
      "path": "qwen38_27b_bench/runtime-verification.json",
      "sha256": "1d79f3652243ba0786a58e7d94700ceab46dcac5c15f07102b3329c4fa2df502"
    },
    {
      "path": "e9_bench/results/run_20260929_163156/environment.txt",
      "sha256": "5bc83b7a204780134fe66edd39d000a3cd4ba4e2627f75eb5c6c6f8d456ceb76"
    },
    {
      "path": "tools/engine-e9/Package.swift",
      "sha256": "c4f31f12999f2a1243be6d00e393ddcfff7391961a4cdd0cdf091789fec8c7d2"
    },
    {
      "path": "mlxfast_bench/build_trees.sh",
      "sha256": "b68d8ad11c7622d95bbba65597684bbda1f3dbdd282366a329a9ade2fb64f987"
    },
    {
      "path": "docs/28092026_mlxfast_256k_plan.md",
      "sha256": "630ba9431abda61e2a45ec5ca2c8ec25e06995f40088b1ba5cad9c95843a0d9d"
    },
    {
      "path": "mlxfast_bench/results/c32k_fans/ctl_serial_1.json",
      "sha256": "130c8a4818ca979a6724bb30eca2ef904f1387ccbcdb53ee42b93bd9b1e15fdd"
    },
    {
      "path": "mlxfast_bench/results/c32k_fans/winglock_serial.json",
      "sha256": "0039e94e9a041f5987f699631a4ecb468e65636d426bc183dbcb0594a1b19538"
    },
    {
      "path": "eight_models_bench/results/run_20261002_153149/ling124_2048_1234_mlx-lm_serial_0/ling-adapter.json",
      "sha256": "8611d57154fa96a6373d5ce5b43a9a9340f4ade3eaaa5e1ff9d898d0b375fc95"
    },
    {
      "path": "eight_models_bench/results/run_20261002_153149/ling124_260864_1234_mlx-lm_serial_2/ling-adapter.json",
      "sha256": "8611d57154fa96a6373d5ce5b43a9a9340f4ade3eaaa5e1ff9d898d0b375fc95"
    },
    {
      "path": "eight_models_bench/results/run_20261002_153149/ling124_32768_1234_mlx-lm_serial_1/ling-adapter.json",
      "sha256": "8611d57154fa96a6373d5ce5b43a9a9340f4ade3eaaa5e1ff9d898d0b375fc95"
    },
    {
      "path": "eight_models_bench/results/run_20261002_173233/ling124_260864_1234_mlx-lm_serial_6/ling-adapter.json",
      "sha256": "8611d57154fa96a6373d5ce5b43a9a9340f4ade3eaaa5e1ff9d898d0b375fc95"
    },
    {
      "path": "mlxfast_bench/results/c32k_fans/ctl_dflash10.json",
      "sha256": "4d57d35cc697bb7ce30cf77aff68a8095c4ef3082483b7091059ed7e46075dff"
    },
    {
      "path": "mlxfast_bench/results/c32k_fans/winglock_dfsafe10.json",
      "sha256": "c645da952fdde079a186054fb791e495a001a7bec766275c6c928b69df1ce449"
    },
    {
      "path": "qwen38_variants_bench/results/run_20261004_225650/tf_flash_260864_1234_tensorfold_mtp_1/result.json",
      "sha256": "a0ce820af154ac25fd6f2e2f7a8c2f339c3c851834e3c61d9b662ae1062e6357"
    }
  ],
  "runtimes": [
    {
      "id": "tensorfold065",
      "engine": "TensorFold",
      "version": "0.6.5",
      "revision": "609ca419abecebdc5a059498a613680bd3aa847f",
      "scope": "Final Qwen3.8-Flash-Next TensorFold tests and all new GLM tests.",
      "dependencies": [
        {
          "name": "mlx",
          "version": "0.32.2",
          "revision": null
        },
        {
          "name": "mlx-lm",
          "version": "0.31.3",
          "revision": null
        }
      ],
      "notes": [
        "The matrix arms set TF_FLASH_DENSE=matrix.",
        "The GLM eager_load arms add a local array-read boundary before GPU layout operations.",
        "The eager loader restores its hooks before each request."
      ],
      "sources": [
        "qwen38_variants_bench/tensorfold-source.json",
        "qwen38_variants_bench/tensorfold-integrity.json",
        "glm53_tensorfold065_bench/runtime-integrity-start.json",
        "glm53_tensorfold065_bench/eager_weights.py"
      ],
      "url": "https://github.com/ashhart/TensorFold/commit/609ca419abecebdc5a059498a613680bd3aa847f",
      "record_count": 44,
      "campaigns": [
        "qwen38",
        "glm065"
      ]
    },
    {
      "id": "mtplx2122",
      "engine": "MTPLX",
      "version": "2.12.2",
      "revision": null,
      "scope": "Qwen3.8-27B and Flash-Next Bare Speed and Optimized Speed tests.",
      "dependencies": [
        {
          "name": "mlx",
          "version": "0.32.2",
          "revision": null
        },
        {
          "name": "mlx-lm",
          "version": "0.31.3",
          "revision": null
        }
      ],
      "notes": [
        "Serial and MTP arms use the same target package.",
        "MTP can combine the native draft head with context-copy proposals."
      ],
      "sources": [
        "qwen38_variants_bench/runtime-integrity.json"
      ],
      "url": null,
      "record_count": 23,
      "campaigns": [
        "qwen38"
      ]
    },
    {
      "id": "tensorfold0363",
      "engine": "TensorFold",
      "version": "0.3.6.3",
      "revision": "191188075bca56a7c71074a79375eb4c1cb22e1c",
      "scope": "Early Nemotron and Qwen3.8-Flash-Next tests.",
      "dependencies": [
        {
          "name": "mlx",
          "version": "0.32.2",
          "revision": null
        },
        {
          "name": "mlx-lm",
          "version": "0.31.3",
          "revision": null
        }
      ],
      "notes": [
        "These tests use the temperature-1 protocol.",
        "The accelerated arms include native MTP and context copies."
      ],
      "sources": [
        "docs/20260929_tensorfold_nemotron_256k.md",
        "tensorfold_bench/requirements.lock.txt",
        "flashnext_bench/requirements.lock.txt"
      ],
      "url": "https://github.com/ashhart/TensorFold/commit/191188075bca56a7c71074a79375eb4c1cb22e1c",
      "record_count": 21,
      "campaigns": [
        "flash_pilot",
        "nemotron"
      ]
    },
    {
      "id": "tensorfold0363_glm",
      "engine": "TensorFold",
      "version": "0.3.6.3 + local compatibility patch",
      "revision": "191188075bca56a7c71074a79375eb4c1cb22e1c",
      "scope": "Earlier GLM standard 4-bit, oQ4, and oQ2 tests.",
      "dependencies": [
        {
          "name": "mlx",
          "version": "0.32.2",
          "revision": null
        },
        {
          "name": "mlx-lm",
          "version": "0.31.3",
          "revision": null
        }
      ],
      "notes": [
        "The patch adjusts affine-width acceptance and tokenizer setup.",
        "It changes no numerical kernel or model weight.",
        "Patch SHA256: 13667562e648fae56db51a20b2633a7653cb5655bdc9709ce84c65a3079d3d66."
      ],
      "sources": [
        "glm53_bench/runtime-provenance.json",
        "glm53_bench/runtime.patch",
        "docs/20260930_glm53_flash_variants.md"
      ],
      "url": "https://github.com/ashhart/TensorFold/commit/191188075bca56a7c71074a79375eb4c1cb22e1c",
      "record_count": 18,
      "campaigns": [
        "glm"
      ]
    },
    {
      "id": "mlx_lm_eight",
      "engine": "mlx-lm",
      "version": "0.32.1.dev4+g53b9af378",
      "revision": "53b9af378278d6dd5dacac447edeb0f523b16253",
      "scope": "Spark, Qwen3.5-2B, Gemma baseline, MiMo baseline, KAT, and the first failed Ling load.",
      "dependencies": [
        {
          "name": "mlx",
          "version": "0.32.2",
          "revision": null
        }
      ],
      "notes": [
        "This group includes the Qwen KV cache variants and prefill-pause serial requests.",
        "The first Ling load failed before the local adapter was added."
      ],
      "sources": [
        "eight_models_bench/runtime-integrity.json",
        "eight_models_bench/worker.py"
      ],
      "url": "https://github.com/ml-explore/mlx-lm/commit/53b9af378278d6dd5dacac447edeb0f523b16253",
      "record_count": 44,
      "campaigns": [
        "eight"
      ]
    },
    {
      "id": "mlx_lm_north",
      "engine": "mlx-lm + mlx-vlm model",
      "version": "0.32.1.dev4+g53b9af378",
      "revision": "53b9af378278d6dd5dacac447edeb0f523b16253",
      "scope": "All North-Mini-Code requests.",
      "dependencies": [
        {
          "name": "mlx",
          "version": "0.32.2",
          "revision": null
        },
        {
          "name": "mlx-vlm",
          "version": "0.7.4",
          "revision": "66e68ce36816cb5134acba3b0d72c6271f35ba24"
        }
      ],
      "notes": [
        "mlx-lm supplies the generation loop.",
        "The mlx-vlm Cohere model uses a local logits-only return adapter."
      ],
      "sources": [
        "eight_models_bench/runtime-integrity.json",
        "eight_models_bench/worker.py"
      ],
      "url": "https://github.com/ml-explore/mlx-lm/commit/53b9af378278d6dd5dacac447edeb0f523b16253",
      "record_count": 7,
      "campaigns": [
        "eight"
      ]
    },
    {
      "id": "mlx_lm_ling",
      "engine": "mlx-lm + local Ling adapter",
      "version": "0.32.1.dev4+g53b9af378",
      "revision": "53b9af378278d6dd5dacac447edeb0f523b16253",
      "scope": "Ling requests after the initial weight-validation failure.",
      "dependencies": [
        {
          "name": "mlx",
          "version": "0.32.2",
          "revision": null
        }
      ],
      "notes": [
        "The adapter comes from the TensorFold model package, with local decay, normalization, and beta-precision corrections.",
        "The TensorFold inference server does not execute these requests.",
        "Adapter helper SHA256: 2cab71f8ad08c24865c518389db9e395f35d41f83971737d4dad5ff6f94a86d6."
      ],
      "sources": [
        "eight_models_bench/runtime-integrity.json",
        "eight_models_bench/ling_adapter.py",
        "eight_models_bench/ling/ADAPTER.md"
      ],
      "url": "https://github.com/ml-explore/mlx-lm/commit/53b9af378278d6dd5dacac447edeb0f523b16253",
      "record_count": 4,
      "campaigns": [
        "eight"
      ]
    },
    {
      "id": "mlx_vlm_gemma",
      "engine": "mlx-vlm",
      "version": "0.7.4",
      "revision": "66e68ce36816cb5134acba3b0d72c6271f35ba24",
      "scope": "Gemma assistant tests with gemma-vlm_serial and gemma-vlm_mtp modes.",
      "dependencies": [
        {
          "name": "mlx",
          "version": "0.32.2",
          "revision": null
        },
        {
          "name": "mlx-lm",
          "version": "0.32.1.dev4+g53b9af378",
          "revision": "53b9af378278d6dd5dacac447edeb0f523b16253"
        }
      ],
      "notes": [
        "These short accelerator tests use the direct mlx-vlm bridge.",
        "The earlier Gemma baseline requests use mlx-lm."
      ],
      "sources": [
        "eight_models_bench/runtime-integrity.json",
        "eight_models_bench/accel_engines.py"
      ],
      "url": "https://github.com/Blaizzy/mlx-vlm/commit/66e68ce36816cb5134acba3b0d72c6271f35ba24",
      "record_count": 6,
      "campaigns": [
        "eight"
      ]
    },
    {
      "id": "mtplx2120_mimo",
      "engine": "MTPLX",
      "version": "2.12.0",
      "revision": null,
      "scope": "MiMo 9B MTPLX serial and MTP accelerator tests.",
      "dependencies": [
        {
          "name": "mlx",
          "version": "0.32.3",
          "revision": null
        },
        {
          "name": "mlx-lm",
          "version": "0.31.3",
          "revision": null
        }
      ],
      "notes": [
        "This separate environment uses the 6-bit target and the borrowed Qwen draft head.",
        "These tests differ from the mixed-precision MiMo baseline."
      ],
      "sources": [
        "eight_models_bench/mtplx-runtime-integrity.json",
        "eight_models_bench/accel_engines.py"
      ],
      "url": null,
      "record_count": 2,
      "campaigns": [
        "eight"
      ]
    },
    {
      "id": "ollama_mlx_metal_v3",
      "engine": "Ollama / mlx_metal_v3",
      "version": "0.35.0",
      "revision": null,
      "scope": "qwen3.8:27b-mlx requests with the mlx_metal_v3 backend.",
      "dependencies": [
        {
          "name": "MLX backend",
          "version": "Bundled with Ollama 0.35.0",
          "revision": null
        }
      ],
      "notes": [
        "Both Metal backends come from the same verified Ollama release archive.",
        "Archive SHA256: 2608dbb0a0f0136a198db9d48b4f74ece55f452314a39452fca35b7cf20c2589."
      ],
      "sources": [
        "qwen38_27b_bench/runtime-version.json",
        "qwen38_27b_bench/runtime-verification.json"
      ],
      "url": "https://github.com/ollama/ollama/releases/tag/v0.35.0",
      "record_count": 5,
      "campaigns": [
        "ollama"
      ]
    },
    {
      "id": "ollama_mlx_metal_v4",
      "engine": "Ollama / mlx_metal_v4",
      "version": "0.35.0",
      "revision": null,
      "scope": "qwen3.8:27b-mlx requests with the mlx_metal_v4 backend.",
      "dependencies": [
        {
          "name": "MLX backend",
          "version": "Bundled with Ollama 0.35.0",
          "revision": null
        }
      ],
      "notes": [
        "Both Metal backends come from the same verified Ollama release archive.",
        "Archive SHA256: 2608dbb0a0f0136a198db9d48b4f74ece55f452314a39452fca35b7cf20c2589."
      ],
      "sources": [
        "qwen38_27b_bench/runtime-version.json",
        "qwen38_27b_bench/runtime-verification.json"
      ],
      "url": "https://github.com/ollama/ollama/releases/tag/v0.35.0",
      "record_count": 7,
      "campaigns": [
        "ollama"
      ]
    },
    {
      "id": "engine_e9",
      "engine": "ENGINE",
      "version": "Git revision",
      "revision": "f5078d82065f83634383f5cc4735c81f152d3ba9",
      "scope": "All four Flash-Next E9 requests, including the cached replay.",
      "dependencies": [
        {
          "name": "mlx-swift",
          "version": "Vendored source",
          "revision": "df1fdc5f7821a1fabe921fdefbc42ac74dcfb6bc"
        },
        {
          "name": "mlx-swift-lm",
          "version": "Vendored source",
          "revision": "bc1c0ee67d15798343be17c9f8f61f7c0d977149"
        },
        {
          "name": "Swift",
          "version": "6.3.2",
          "revision": null
        }
      ],
      "notes": [
        "Native MTP uses depth 3 and the temperature-1 protocol.",
        "The source commit identifies the engine build."
      ],
      "sources": [
        "e9_bench/results/run_20260929_163156/environment.txt",
        "tools/engine-e9/Package.swift"
      ],
      "url": "https://github.com/KrakowiakK/ENGINE/commit/f5078d82065f83634383f5cc4735c81f152d3ba9",
      "record_count": 4,
      "campaigns": [
        "e9"
      ]
    },
    {
      "id": "mlxfast_control",
      "engine": "mlx.fast Swift/Metal · Reference",
      "version": "Git revision",
      "revision": "88cf569e885019ffe637192efcbda4aaf2c32f1c",
      "scope": "Reference serial and DFlash-10 performance arms at 31,958 input tokens.",
      "dependencies": [],
      "notes": [
        "The reference build combines its source revision with the trusted harness at 831fae740de35a106e85768d8b0534b4af422b13.",
        "DFlash depth 10 is the M3 compatibility configuration.",
        "Worker SHA256: 4e5c6324b37d9b2a3bb256e39720a9959d6efdadbec67025f8eefde5ff57db7f."
      ],
      "sources": [
        "mlxfast_bench/results/c32k_fans/ctl_serial_1.json",
        "mlxfast_bench/build_trees.sh",
        "docs/28092026_mlxfast_256k_plan.md"
      ],
      "url": null,
      "record_count": 2,
      "campaigns": [
        "mlxfast"
      ]
    },
    {
      "id": "mlxfast_winglock",
      "engine": "mlx.fast Swift/Metal · Winglock",
      "version": "Git revision",
      "revision": "831fae740de35a106e85768d8b0534b4af422b13",
      "scope": "Winglock serial and DFlash-10 performance arms at 31,958 input tokens.",
      "dependencies": [],
      "notes": [
        "The Winglock build uses the recorded source revision.",
        "DFlash depth 10 is the M3 compatibility configuration.",
        "Worker SHA256: 1db295e0d6c8653b3adc1dc079d5db4b702d628b7235d7782a73a8bd367a4901."
      ],
      "sources": [
        "mlxfast_bench/results/c32k_fans/winglock_serial.json",
        "mlxfast_bench/build_trees.sh",
        "docs/28092026_mlxfast_256k_plan.md"
      ],
      "url": null,
      "record_count": 2,
      "campaigns": [
        "mlxfast"
      ]
    }
  ],
  "model_revisions": [
    {
      "key": "27b_bare",
      "repo": "Youssofal/Qwen3.8-27B-MTPLX-Bare-Speed",
      "revision": "b59d7002368575f08a58ba0d26686b53a9c162d6"
    },
    {
      "key": "27b_speed",
      "repo": "Youssofal/Qwen3.8-27B-MTPLX-Optimized-Speed",
      "revision": "1d5087d2062c02b279180a53e4016cf9cd7a3d7e"
    },
    {
      "key": "flash_bare",
      "repo": "Youssofal/Qwen3.8-Flash-Next-MTPLX-Bare-Speed",
      "revision": "4285a4a04041511bf6581f543fdbfd154c93a608"
    },
    {
      "key": "flash_speed",
      "repo": "Youssofal/Qwen3.8-Flash-Next-MTPLX-Optimized-Speed",
      "revision": "34ae535f2c817737a9c73c364fd6f20629f9a8f7"
    },
    {
      "key": "spark17",
      "repo": "XHToken/Spark-X2.5-1.7B",
      "revision": "14d6e83c13c7add2b62a7c39b2131f4ed1cddcf8",
      "quant": "bf16"
    },
    {
      "key": "gemma26",
      "repo": "mlx-community/gemma-4-26B-A4B-it-qat-4bit",
      "revision": "0e3cbab38ce568cf6e23543010d08d03b731910c",
      "quant": "qat4"
    },
    {
      "key": "qwen2",
      "repo": "mlx-community/Qwen3.5-2B-4bit",
      "revision": "674aaa7240b91e8012fcad5d791b7dfe5ba90207",
      "quant": "4bit"
    },
    {
      "key": "spark4",
      "repo": "abenzerps/Spark-X2.5-4B-MLX-8bit",
      "revision": "8dedb0ea144edd63b0b9b72103bc012b422466eb",
      "quant": "8bit"
    },
    {
      "key": "mimo9",
      "repo": "DJLougen/MiMo-V2.6-Distill-Qwen-9B-MLX-mixed-4-6",
      "revision": "90f7021d9decf475dbf7c708d4918591595982bb",
      "quant": "mixed4_6"
    },
    {
      "key": "kat35",
      "repo": "mlx-community/KAT-Coder-V2.5-Dev-OptiQ-4bit",
      "revision": "57c6e8372737f8a04aca1c210c926b8aa29c802e",
      "quant": "optiq4"
    },
    {
      "key": "north30",
      "repo": "mlx-community/North-Mini-Code-1.0-4bit",
      "revision": "dfbe084dfa26e241345af99ca32848f38fd865f9",
      "quant": "4bit"
    },
    {
      "key": "ling124",
      "repo": "TensorFold/Ling-3.0-flash-MLX-4bit",
      "revision": "cae62774492b589eea4aab2d6d36fb167a67d90c",
      "quant": "4bit"
    },
    {
      "key": "gemma_assistant",
      "repo": "mlx-community/gemma-4-26B-A4B-it-qat-assistant-4bit",
      "revision": "bb94eae1b70a80dac16cbf959bb4b7d56bd1fb8c",
      "quant": "qat4"
    },
    {
      "key": "glm_4bit",
      "repo": "Vontra/GLM-5.3-Flash-MLX-4bit-MTP",
      "revision": "76add2a341a1cd90ad0e86bb69839ea9c35827c6"
    },
    {
      "key": "glm_oq4",
      "repo": "Vontra/GLM-5.3-Flash-MLX-oQ4-MTP",
      "revision": "9938ce0dbae9d9cd0f5c17bbd57559d9d5a23196"
    },
    {
      "key": "glm_oq2",
      "repo": "Vontra/GLM-5.3-Flash-MLX-oQ2-MTP",
      "revision": "5da47563173b3c0b5fb4065ab7676477a7d50c44"
    },
    {
      "key": "tf_flash",
      "repo": "Vontra/Qwen3.8-Flash-Next-MLX-4bit-MTP",
      "revision": "dadefa8066e3be900a0d148d0f5a2f4eb1cf6534"
    }
  ],
  "weight_formats": [
    {
      "key": "glm_4bit",
      "default": {
        "bits": 4,
        "group_size": 64,
        "mode": "affine"
      },
      "configured_module_overrides": [],
      "scope": "The counts describe configuration entries. They do not count weight tensors or parameter values."
    },
    {
      "key": "glm_oq4",
      "default": {
        "bits": 4,
        "group_size": 64,
        "mode": "affine"
      },
      "configured_module_overrides": [
        {
          "bits": 5,
          "group_size": 64,
          "modules": 76
        },
        {
          "bits": 6,
          "group_size": 64,
          "modules": 17
        },
        {
          "bits": 8,
          "group_size": 64,
          "modules": 461
        }
      ],
      "scope": "The counts describe configuration entries. They do not count weight tensors or parameter values."
    },
    {
      "key": "glm_oq2",
      "default": {
        "bits": 2,
        "group_size": 64,
        "mode": "affine"
      },
      "configured_module_overrides": [
        {
          "bits": 4,
          "group_size": 64,
          "modules": 3
        },
        {
          "bits": 8,
          "group_size": 64,
          "modules": 554
        }
      ],
      "scope": "The counts describe configuration entries. They do not count weight tensors or parameter values."
    }
  ],
  "measurements": [
    {
      "campaign": "qwen38",
      "source": "qwen38_variants_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "qwen38_variants_bench/results/run_20261004_185731/27b_bare_2048_1234_mtplx_mtp_1",
      "model": "27b_bare",
      "mode": "mtplx_mtp",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 61.99097593285372,
      "metric": "Output token slope after token 32",
      "runtime_tps": 63.18953123750971,
      "ttft_seconds": 4.694591667037457,
      "gpu_mhz": 1377,
      "peak_load": 2.4267578125,
      "recall": 8,
      "qualified": false,
      "failed_checks": [
        "matched_serial_exactness"
      ],
      "quality_pass": true,
      "exact_pair": false,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "8f10b1022c43155b49d16021cc59817add0367496ccaffdcfaf4daada06197ff",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "The first eight lines contain the correct city-code pairs.",
          "The later text addresses the archive task.",
          "The output ends at the 1024-token limit.",
          "This review does not validate each claim in the later text."
        ]
      },
      "runtime_id": "mtplx2122"
    },
    {
      "campaign": "qwen38",
      "source": "qwen38_variants_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "qwen38_variants_bench/results/run_20261004_185731/27b_bare_2048_1234_mtplx_serial_0",
      "model": "27b_bare",
      "mode": "mtplx_serial",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 32.41352179316568,
      "metric": "Output token slope after token 32",
      "runtime_tps": 32.45984681602271,
      "ttft_seconds": 4.607190417125821,
      "gpu_mhz": 1380,
      "peak_load": 1.7216796875,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "18180bbf1cc4e6ce8ef0e5de70b07217b3c3f389dcd57d5befdef85ffc034658",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "The first eight lines contain the correct city-code pairs.",
          "The later text addresses the archive task.",
          "The output ends at the 1024-token limit.",
          "This review does not validate each claim in the later text."
        ]
      },
      "runtime_id": "mtplx2122"
    },
    {
      "campaign": "qwen38",
      "source": "qwen38_variants_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "qwen38_variants_bench/results/run_20261004_185731/27b_bare_32768_1234_mtplx_mtp_3",
      "model": "27b_bare",
      "mode": "mtplx_mtp",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 50.93284130008731,
      "metric": "Output token slope after token 32",
      "runtime_tps": 50.394915604958086,
      "ttft_seconds": 84.77246291702613,
      "gpu_mhz": 1377.5,
      "peak_load": 2.5615234375,
      "recall": 8,
      "qualified": false,
      "failed_checks": [
        "matched_serial_exactness"
      ],
      "quality_pass": true,
      "exact_pair": false,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "f5fc4ade7a5b840fa725c00bd1851c14d6c1ba5794c7935071afdad165ddd3a4",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "The first eight lines contain the correct city-code pairs.",
          "The later text addresses the archive task.",
          "The output ends at the 1024-token limit.",
          "This review does not validate each claim in the later text."
        ]
      },
      "runtime_id": "mtplx2122"
    },
    {
      "campaign": "qwen38",
      "source": "qwen38_variants_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "qwen38_variants_bench/results/run_20261004_185731/27b_bare_32768_1234_mtplx_serial_2",
      "model": "27b_bare",
      "mode": "mtplx_serial",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 28.07410820202328,
      "metric": "Output token slope after token 32",
      "runtime_tps": 28.08893350811893,
      "ttft_seconds": 82.73025908297859,
      "gpu_mhz": 1380,
      "peak_load": 2.8251953125,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "bd8b0d530f3ef7bc8d902c9576feb8af107837ab4fd76950e2077a6815394210",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "The first eight lines contain the correct city-code pairs.",
          "The later text addresses the archive task.",
          "The output ends at the 1024-token limit.",
          "This review does not validate each claim in the later text."
        ]
      },
      "runtime_id": "mtplx2122"
    },
    {
      "campaign": "qwen38",
      "source": "qwen38_variants_bench/all-results.json",
      "status": "incomplete",
      "error": "Prefill exceeded 3600 seconds",
      "id": "qwen38_variants_bench/results/run_20261004_185731/27b_bare_260864_1234_mtplx_serial_4",
      "model": "27b_bare",
      "mode": "mtplx_serial",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": null,
      "output_tokens": null,
      "cached_tokens": null,
      "decode_tps": null,
      "metric": "Output token slope after token 32",
      "runtime_tps": null,
      "ttft_seconds": null,
      "gpu_mhz": null,
      "peak_load": 12.3603515625,
      "recall": null,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": null,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": null,
      "abort_reasons": [
        "Prefill exceeded 3600 seconds"
      ],
      "stop_reason": "Prefill exceeded 3600 seconds",
      "manual_review": null,
      "runtime_id": "mtplx2122"
    },
    {
      "campaign": "qwen38",
      "source": "qwen38_variants_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "qwen38_variants_bench/results/run_20261004_202905/27b_speed_2048_1234_mtplx_mtp_1",
      "model": "27b_speed",
      "mode": "mtplx_mtp",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 55.47231951527046,
      "metric": "Output token slope after token 32",
      "runtime_tps": 57.094035822020885,
      "ttft_seconds": 4.76269037486054,
      "gpu_mhz": 1377.5,
      "peak_load": 2.52685546875,
      "recall": 8,
      "qualified": false,
      "failed_checks": [
        "matched_serial_exactness"
      ],
      "quality_pass": true,
      "exact_pair": false,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "9c4c196d7297109645bba56bc87a73c934044de92293fe6434c00e686680a625",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "The first eight lines contain the correct city-code pairs.",
          "The later text addresses the archive task.",
          "The output ends at the 1024-token limit.",
          "This review does not validate each claim in the later text."
        ]
      },
      "runtime_id": "mtplx2122"
    },
    {
      "campaign": "qwen38",
      "source": "qwen38_variants_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "qwen38_variants_bench/results/run_20261004_202905/27b_speed_2048_1234_mtplx_serial_0",
      "model": "27b_speed",
      "mode": "mtplx_serial",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 27.44046019717171,
      "metric": "Output token slope after token 32",
      "runtime_tps": 27.48146773288366,
      "ttft_seconds": 4.667571707861498,
      "gpu_mhz": 1380,
      "peak_load": 3.916015625,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "3d1c46eeb9423f620c4753a756c02bd58449fa26ec07846be2506d3a35005de2",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "The first eight lines contain the correct city-code pairs.",
          "The later text addresses the archive task.",
          "The output ends at the 1024-token limit.",
          "This review does not validate each claim in the later text."
        ]
      },
      "runtime_id": "mtplx2122"
    },
    {
      "campaign": "qwen38",
      "source": "qwen38_variants_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "qwen38_variants_bench/results/run_20261004_202905/27b_speed_32768_1234_mtplx_mtp_3",
      "model": "27b_speed",
      "mode": "mtplx_mtp",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 43.42155228150668,
      "metric": "Output token slope after token 32",
      "runtime_tps": 44.38341091455755,
      "ttft_seconds": 85.4573814170435,
      "gpu_mhz": 1378,
      "peak_load": 2.76708984375,
      "recall": 8,
      "qualified": false,
      "failed_checks": [
        "matched_serial_exactness"
      ],
      "quality_pass": true,
      "exact_pair": false,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "b8d0e125a8f360d881832d9ce8f7292d7aa7e84eb541bb6233bfd76d8c24be3b",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "The first eight lines contain the correct city-code pairs.",
          "The later text addresses the archive task.",
          "The output ends at the 1024-token limit.",
          "This review does not validate each claim in the later text."
        ]
      },
      "runtime_id": "mtplx2122"
    },
    {
      "campaign": "qwen38",
      "source": "qwen38_variants_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "qwen38_variants_bench/results/run_20261004_202905/27b_speed_32768_1234_mtplx_serial_2",
      "model": "27b_speed",
      "mode": "mtplx_serial",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 24.174837777102105,
      "metric": "Output token slope after token 32",
      "runtime_tps": 24.196448960442602,
      "ttft_seconds": 83.41215699980967,
      "gpu_mhz": 1380,
      "peak_load": 2.84375,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "7c17450c260ce39067d45c62f2f0611ebabad4907c8e6c959097c42a97f40197",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "The first eight lines contain the correct city-code pairs.",
          "The later text addresses the archive task.",
          "The spelling example repeats the same name without a visible difference.",
          "The output ends at the 1024-token limit.",
          "This review does not validate each claim in the later text."
        ]
      },
      "runtime_id": "mtplx2122"
    },
    {
      "campaign": "qwen38",
      "source": "qwen38_variants_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "qwen38_variants_bench/results/run_20261004_211626/flash_bare_2048_1234_mtplx_mtp_1",
      "model": "flash_bare",
      "mode": "mtplx_mtp",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 65.09551016519193,
      "metric": "Output token slope after token 32",
      "runtime_tps": 65.78370600080366,
      "ttft_seconds": 1.7637630419339985,
      "gpu_mhz": 1380,
      "peak_load": 2.78076171875,
      "recall": 8,
      "qualified": false,
      "failed_checks": [
        "matched_serial_exactness"
      ],
      "quality_pass": true,
      "exact_pair": false,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "e084b04f94e1f4868de2450baff073f7fecc290f4b3919e0abb65c102f275015",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "The first eight lines contain the correct city-code pairs.",
          "The later text addresses the archive task.",
          "The output ends at the 1024-token limit.",
          "This review does not validate each claim in the later text."
        ]
      },
      "runtime_id": "mtplx2122"
    },
    {
      "campaign": "qwen38",
      "source": "qwen38_variants_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "qwen38_variants_bench/results/run_20261004_211626/flash_bare_2048_1234_mtplx_serial_0",
      "model": "flash_bare",
      "mode": "mtplx_serial",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 40.72815031046409,
      "metric": "Output token slope after token 32",
      "runtime_tps": 40.699343766855684,
      "ttft_seconds": 1.8831269999500364,
      "gpu_mhz": 1373,
      "peak_load": 3.14453125,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "4d743a8f2e7333d2aca1ac558643e95c0865704b13c83ee027b6219efaf66645",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "The first eight lines contain the correct city-code pairs.",
          "The later text addresses the archive task.",
          "The output ends at the 1024-token limit.",
          "This review does not validate each claim in the later text."
        ]
      },
      "runtime_id": "mtplx2122"
    },
    {
      "campaign": "qwen38",
      "source": "qwen38_variants_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "qwen38_variants_bench/results/run_20261004_211626/flash_bare_32768_1234_mtplx_mtp_3",
      "model": "flash_bare",
      "mode": "mtplx_mtp",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 54.59279774852212,
      "metric": "Output token slope after token 32",
      "runtime_tps": 55.69701997959966,
      "ttft_seconds": 36.236676667118445,
      "gpu_mhz": 1380,
      "peak_load": 5.5595703125,
      "recall": 8,
      "qualified": false,
      "failed_checks": [
        "matched_serial_exactness"
      ],
      "quality_pass": true,
      "exact_pair": false,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "486052b7115645da6bc69f20610071c741c50f8f6ea1ac498e8e2eaa6be10c4e",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "The first eight lines contain the correct city-code pairs.",
          "The later text addresses the archive task.",
          "Some later claims about the archive lack evidence.",
          "The output ends at the 1024-token limit."
        ]
      },
      "runtime_id": "mtplx2122"
    },
    {
      "campaign": "qwen38",
      "source": "qwen38_variants_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "qwen38_variants_bench/results/run_20261004_211626/flash_bare_32768_1234_mtplx_serial_2",
      "model": "flash_bare",
      "mode": "mtplx_serial",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 38.063309334799015,
      "metric": "Output token slope after token 32",
      "runtime_tps": 38.156058244398174,
      "ttft_seconds": 34.84539512498304,
      "gpu_mhz": 1373,
      "peak_load": 2.9384765625,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "4f29da7860c021bf0cbcbd0eef5de287f0670680cd7b19ba9bfca85e6d492fa9",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "The first eight lines contain the correct city-code pairs.",
          "The later text addresses the archive task.",
          "The output ends at the 1024-token limit.",
          "This review does not validate each claim in the later text."
        ]
      },
      "runtime_id": "mtplx2122"
    },
    {
      "campaign": "qwen38",
      "source": "qwen38_variants_bench/all-results.json",
      "status": "incomplete",
      "error": "KeyboardInterrupt('signal 15')",
      "id": "qwen38_variants_bench/results/run_20261005_002720/flash_bare_260864_1234_mtplx_mtp_1",
      "model": "flash_bare",
      "mode": "mtplx_mtp",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": null,
      "output_tokens": null,
      "cached_tokens": null,
      "decode_tps": null,
      "metric": "Output token slope after token 32",
      "runtime_tps": null,
      "ttft_seconds": null,
      "gpu_mhz": null,
      "peak_load": 21.076171875,
      "recall": null,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": null,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": null,
      "abort_reasons": [
        "System load exceeded16 twice"
      ],
      "stop_reason": null,
      "manual_review": null,
      "runtime_id": "mtplx2122"
    },
    {
      "campaign": "qwen38",
      "source": "qwen38_variants_bench/all-results.json",
      "status": "incomplete",
      "error": "KeyboardInterrupt('signal 15')",
      "id": "qwen38_variants_bench/results/run_20261004_211626/flash_bare_260864_1234_mtplx_serial_4",
      "model": "flash_bare",
      "mode": "mtplx_serial",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": null,
      "output_tokens": null,
      "cached_tokens": null,
      "decode_tps": null,
      "metric": "Output token slope after token 32",
      "runtime_tps": null,
      "ttft_seconds": null,
      "gpu_mhz": null,
      "peak_load": 13.78662109375,
      "recall": null,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": null,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": null,
      "abort_reasons": [],
      "stop_reason": "System load exceeded the qualification limit; the TensorFold control takes priority",
      "manual_review": null,
      "runtime_id": "mtplx2122"
    },
    {
      "campaign": "qwen38",
      "source": "qwen38_variants_bench/all-results.json",
      "status": "incomplete",
      "error": "RuntimeError('[METAL] Command buffer execution failed: Caused GPU Timeout Error (00000002:kIOGPUCommandBufferCallbackErrorTimeout).')",
      "id": "qwen38_variants_bench/results/run_20261005_001052/flash_bare_260864_1234_mtplx_serial_4",
      "model": "flash_bare",
      "mode": "mtplx_serial",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": null,
      "output_tokens": null,
      "cached_tokens": null,
      "decode_tps": null,
      "metric": "Output token slope after token 32",
      "runtime_tps": null,
      "ttft_seconds": null,
      "gpu_mhz": null,
      "peak_load": null,
      "recall": null,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": null,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": null,
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": null,
      "runtime_id": "mtplx2122"
    },
    {
      "campaign": "qwen38",
      "source": "qwen38_variants_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "qwen38_variants_bench/results/run_20261005_002720/flash_bare_260864_1234_mtplx_serial_0",
      "model": "flash_bare",
      "mode": "mtplx_serial",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 23.345691915633505,
      "metric": "Output token slope after token 32",
      "runtime_tps": 23.51723541242785,
      "ttft_seconds": 2322.446007957915,
      "gpu_mhz": 885,
      "peak_load": 4.556640625,
      "recall": 8,
      "qualified": false,
      "failed_checks": [
        "normal_decode_clock"
      ],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "bd12c413e4d4f0ce24ed069fe6caea9b7c4de355e0567953393efea9ae001456",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "The first eight lines contain the correct city-code pairs.",
          "The later text addresses archive verification and conflict checks.",
          "The output ends at the 1024-token limit.",
          "This review does not validate every claim or quotation in the later text."
        ]
      },
      "runtime_id": "mtplx2122"
    },
    {
      "campaign": "qwen38",
      "source": "qwen38_variants_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "qwen38_variants_bench/results/run_20261005_001052/flash_speed_2048_1234_mtplx_mtp_1",
      "model": "flash_speed",
      "mode": "mtplx_mtp",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 64.50655200373502,
      "metric": "Output token slope after token 32",
      "runtime_tps": 65.65494750384099,
      "ttft_seconds": 1.7913552080281079,
      "gpu_mhz": 1380,
      "peak_load": 2.71142578125,
      "recall": 8,
      "qualified": false,
      "failed_checks": [
        "matched_serial_exactness"
      ],
      "quality_pass": true,
      "exact_pair": false,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "541716df22d855e819bdf854349b9dcb6e68e24f980a412356be8013d412ecd3",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "The first eight lines contain the correct city-code pairs.",
          "The later text addresses the archive task.",
          "The cited Reykjavik context matches the archive.",
          "The output ends at the 1024-token limit.",
          "This review does not validate every claim in the later text."
        ]
      },
      "runtime_id": "mtplx2122"
    },
    {
      "campaign": "qwen38",
      "source": "qwen38_variants_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "qwen38_variants_bench/results/run_20261005_001052/flash_speed_2048_1234_mtplx_serial_0",
      "model": "flash_speed",
      "mode": "mtplx_serial",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 37.88022678692755,
      "metric": "Output token slope after token 32",
      "runtime_tps": 37.908576552027576,
      "ttft_seconds": 1.779785375110805,
      "gpu_mhz": 1373,
      "peak_load": 2.6318359375,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "a2e68eb4edc04fdf110a1d4c32f3551bcb6cdc111997be54a93428a8ca77e284",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "The first eight lines contain the correct city-code pairs.",
          "The later text addresses the archive task.",
          "The archive body has the partial final phrase that the answer describes.",
          "The output ends at the 1024-token limit.",
          "This review does not validate every claim in the later text."
        ]
      },
      "runtime_id": "mtplx2122"
    },
    {
      "campaign": "qwen38",
      "source": "qwen38_variants_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "qwen38_variants_bench/results/run_20261005_001052/flash_speed_32768_1234_mtplx_mtp_3",
      "model": "flash_speed",
      "mode": "mtplx_mtp",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 67.10005130704316,
      "metric": "Output token slope after token 32",
      "runtime_tps": 68.543892633603,
      "ttft_seconds": 37.29132608301006,
      "gpu_mhz": 1380,
      "peak_load": 3.1318359375,
      "recall": 8,
      "qualified": false,
      "failed_checks": [
        "matched_serial_exactness"
      ],
      "quality_pass": true,
      "exact_pair": false,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "76af15a6bc4589cbffab385dfaf569f2b266640fc22acfc24f4804a04a8462a5",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "The first eight lines contain the correct city-code pairs.",
          "The later text addresses the archive task.",
          "The guide repeats the eight records once as extraction results.",
          "The output ends at the 1024-token limit.",
          "This review does not validate every claim in the later text."
        ]
      },
      "runtime_id": "mtplx2122"
    },
    {
      "campaign": "qwen38",
      "source": "qwen38_variants_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "qwen38_variants_bench/results/run_20261005_001052/flash_speed_32768_1234_mtplx_serial_2",
      "model": "flash_speed",
      "mode": "mtplx_serial",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 35.79252873877767,
      "metric": "Output token slope after token 32",
      "runtime_tps": 35.811251884297604,
      "ttft_seconds": 35.64438012498431,
      "gpu_mhz": 1373,
      "peak_load": 2.53955078125,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "35a0ed81c6c64de60fd5c29488e7180fe5b0d013581d32e49f93260dec78d2df",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "The first eight lines contain the correct city-code pairs.",
          "The later text addresses the archive task.",
          "The output ends at the 1024-token limit.",
          "This review does not validate every claim in the later text."
        ]
      },
      "runtime_id": "mtplx2122"
    },
    {
      "campaign": "qwen38",
      "source": "qwen38_variants_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "qwen38_variants_bench/results/run_20261005_014554/flash_speed_260864_1234_mtplx_mtp_1",
      "model": "flash_speed",
      "mode": "mtplx_mtp",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 33.60034818160365,
      "metric": "Output token slope after token 32",
      "runtime_tps": 34.87708033899879,
      "ttft_seconds": 2524.6159641251434,
      "gpu_mhz": 539,
      "peak_load": 9.9609375,
      "recall": 8,
      "qualified": false,
      "failed_checks": [
        "normal_decode_clock",
        "load",
        "matched_serial_exactness"
      ],
      "quality_pass": true,
      "exact_pair": false,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "4a85f8b66fd8bd8ffd3d64dbb988366a8b932fb6484bcc9153694100cde1c253",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "The first eight lines contain the correct city-code pairs.",
          "The later text addresses archive verification and data extraction.",
          "The output ends at the 1024-token limit.",
          "This review does not validate every claim or quotation in the later text."
        ]
      },
      "runtime_id": "mtplx2122"
    },
    {
      "campaign": "qwen38",
      "source": "qwen38_variants_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "qwen38_variants_bench/results/run_20261005_014554/flash_speed_260864_1234_mtplx_serial_0",
      "model": "flash_speed",
      "mode": "mtplx_serial",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 21.578208064023624,
      "metric": "Output token slope after token 32",
      "runtime_tps": 21.759517837615636,
      "ttft_seconds": 2382.918017125223,
      "gpu_mhz": 853.5,
      "peak_load": 7.2451171875,
      "recall": 8,
      "qualified": false,
      "failed_checks": [
        "normal_decode_clock"
      ],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "3e86f29cd2afd9561ed67907052c352b8aa889a014d68ced2d34547f282530d4",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "The first eight lines contain the correct city-code pairs.",
          "The later text addresses archive verification.",
          "The guide says a change in descriptive text indicates a security breach. This test provides no evidence for that claim.",
          "The output ends at the 1024-token limit.",
          "This review does not validate every claim or quotation in the later text."
        ]
      },
      "runtime_id": "mtplx2122"
    },
    {
      "campaign": "qwen38",
      "source": "qwen38_variants_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "qwen38_variants_bench/results/run_20261004_215711/tf_flash_2048_1234_tensorfold_mtp_1",
      "model": "tf_flash",
      "mode": "tensorfold_mtp",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 111.89371784550616,
      "metric": "Output token slope after token 32",
      "runtime_tps": 112.89688864055132,
      "ttft_seconds": 2.5277755421120673,
      "gpu_mhz": 1380,
      "peak_load": 3.38818359375,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": true,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "5b024c979773d0d5aad6b1a863ad439d6d793c3cce1180ea1c072c24c02f5837",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "The first eight lines contain the correct city-code pairs.",
          "The later text addresses the archive task.",
          "The output ends at the 1024-token limit.",
          "This review does not validate every claim in the later text."
        ]
      },
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "qwen38",
      "source": "qwen38_variants_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "qwen38_variants_bench/results/run_20261004_215711/tf_flash_2048_1234_tensorfold_serial_0",
      "model": "tf_flash",
      "mode": "tensorfold_serial",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 84.78425073386157,
      "metric": "Output token slope after token 32",
      "runtime_tps": 84.41783187588159,
      "ttft_seconds": 2.4975239171180874,
      "gpu_mhz": 1380,
      "peak_load": 2.7353515625,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "5b024c979773d0d5aad6b1a863ad439d6d793c3cce1180ea1c072c24c02f5837",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "The first eight lines contain the correct city-code pairs.",
          "The later text addresses the archive task.",
          "The output ends at the 1024-token limit.",
          "This review does not validate every claim in the later text."
        ]
      },
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "qwen38",
      "source": "qwen38_variants_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "qwen38_variants_bench/results/run_20261004_233659/tf_flash_32768_1234_tensorfold_mtp_1",
      "model": "tf_flash",
      "mode": "tensorfold_matrix_mtp",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 101.23553740764268,
      "metric": "Output token slope after token 32",
      "runtime_tps": 101.96066979311941,
      "ttft_seconds": 27.766451542032883,
      "gpu_mhz": 1380,
      "peak_load": 3.1865234375,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": true,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "b7befef80ada4415ba08d994e2e6204cbc119b3b811e1a16048b9d4b369e595b",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "The output bytes equal the manually reviewed matrix serial output.",
          "Every output token matches that serial control.",
          "All eight retrieval checks passed.",
          "This comparison does not validate every claim in the later text."
        ]
      },
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "qwen38",
      "source": "qwen38_variants_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "qwen38_variants_bench/results/run_20261004_233659/tf_flash_32768_1234_tensorfold_serial_0",
      "model": "tf_flash",
      "mode": "tensorfold_matrix_serial",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 78.05236474118176,
      "metric": "Output token slope after token 32",
      "runtime_tps": 77.99116437130368,
      "ttft_seconds": 27.816705167060718,
      "gpu_mhz": 1380,
      "peak_load": 2.63623046875,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "b7befef80ada4415ba08d994e2e6204cbc119b3b811e1a16048b9d4b369e595b",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "The first eight lines contain the correct city-code pairs.",
          "The later text addresses the archive task.",
          "The output ends at the 1024-token limit.",
          "This review does not validate every claim in the later text."
        ]
      },
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "qwen38",
      "source": "qwen38_variants_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "qwen38_variants_bench/results/run_20261004_215711/tf_flash_32768_1234_tensorfold_mtp_3",
      "model": "tf_flash",
      "mode": "tensorfold_mtp",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 113.69602640197563,
      "metric": "Output token slope after token 32",
      "runtime_tps": 116.546080574982,
      "ttft_seconds": 27.834660582942888,
      "gpu_mhz": 1380,
      "peak_load": 2.55517578125,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": true,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "99cd62cdf47faf2fb85b3e71a3b8be81fd9390899cec2f547c2034a925bfa4cc",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "The first eight lines contain the correct city-code pairs.",
          "The later text addresses the archive task.",
          "The output ends at the 1024-token limit.",
          "This review does not validate every claim in the later text."
        ]
      },
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "qwen38",
      "source": "qwen38_variants_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "qwen38_variants_bench/results/run_20261004_215711/tf_flash_32768_1234_tensorfold_serial_2",
      "model": "tf_flash",
      "mode": "tensorfold_serial",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 82.18094437331489,
      "metric": "Output token slope after token 32",
      "runtime_tps": 82.10458887462593,
      "ttft_seconds": 27.77728333394043,
      "gpu_mhz": 1380,
      "peak_load": 2.041015625,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "99cd62cdf47faf2fb85b3e71a3b8be81fd9390899cec2f547c2034a925bfa4cc",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "The first eight lines contain the correct city-code pairs.",
          "The later text addresses the archive task.",
          "The output ends at the 1024-token limit.",
          "This review does not validate every claim in the later text."
        ]
      },
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "qwen38",
      "source": "qwen38_variants_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "qwen38_variants_bench/results/run_20261004_225650/tf_flash_260864_1234_tensorfold_mtp_1",
      "model": "tf_flash",
      "mode": "tensorfold_cooled_prefill_mtp",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 93.24692624799974,
      "metric": "Output token slope after token 32",
      "runtime_tps": 94.50178805126579,
      "ttft_seconds": 735.4590765829198,
      "gpu_mhz": 1379,
      "peak_load": 4.16845703125,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": true,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 489.3494621249847,
      "tokens_before_cool": 1,
      "prefill_pause_seconds": 0,
      "token_sha256": "7d63f38b5fcfc7b5bff58f3d4288a8c7751f35b0b69d138ee0cf6b9720f60f61",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "The output bytes equal the manually reviewed serial output.",
          "Every output token matches the cold MTP reference and the cooled serial control.",
          "All eight retrieval checks passed.",
          "This comparison does not validate every claim in the later text."
        ]
      },
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "qwen38",
      "source": "qwen38_variants_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "qwen38_variants_bench/results/run_20261004_225650/tf_flash_260864_1234_tensorfold_serial_0",
      "model": "tf_flash",
      "mode": "tensorfold_cooled_prefill_serial",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 72.05721484411498,
      "metric": "Output token slope after token 32",
      "runtime_tps": 71.69646094384025,
      "ttft_seconds": 732.93009966705,
      "gpu_mhz": 1380,
      "peak_load": 4.20263671875,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 489.7753640830051,
      "tokens_before_cool": 1,
      "prefill_pause_seconds": 0,
      "token_sha256": "7d63f38b5fcfc7b5bff58f3d4288a8c7751f35b0b69d138ee0cf6b9720f60f61",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "The first eight lines contain the correct city-code pairs.",
          "The later text addresses the archive task.",
          "The output ends at the 1024-token limit.",
          "Every output token matches the cold serial reference.",
          "This review does not validate every claim in the later text."
        ]
      },
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "qwen38",
      "source": "qwen38_variants_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "qwen38_variants_bench/results/run_20261004_215711/tf_flash_260864_1234_tensorfold_mtp_5",
      "model": "tf_flash",
      "mode": "tensorfold_mtp",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 28.918067680424997,
      "metric": "Output token slope after token 32",
      "runtime_tps": 28.93305257008736,
      "ttft_seconds": 410.49917908292264,
      "gpu_mhz": 338,
      "peak_load": 4.0439453125,
      "recall": 8,
      "qualified": false,
      "failed_checks": [
        "normal_decode_clock",
        "decode_clamp"
      ],
      "quality_pass": true,
      "exact_pair": true,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "7d63f38b5fcfc7b5bff58f3d4288a8c7751f35b0b69d138ee0cf6b9720f60f61",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "The first eight lines contain the correct city-code pairs.",
          "The later text addresses the archive task.",
          "The output ends at the 1024-token limit.",
          "This review does not validate every claim in the later text."
        ]
      },
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "qwen38",
      "source": "qwen38_variants_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "qwen38_variants_bench/results/run_20261004_215711/tf_flash_260864_1234_tensorfold_serial_4",
      "model": "tf_flash",
      "mode": "tensorfold_serial",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 44.831823521299725,
      "metric": "Output token slope after token 32",
      "runtime_tps": 45.229412600884835,
      "ttft_seconds": 259.29116729204543,
      "gpu_mhz": 520.5,
      "peak_load": 3.68310546875,
      "recall": 8,
      "qualified": false,
      "failed_checks": [
        "normal_decode_clock"
      ],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "7d63f38b5fcfc7b5bff58f3d4288a8c7751f35b0b69d138ee0cf6b9720f60f61",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "The first eight lines contain the correct city-code pairs.",
          "The later text addresses the archive task.",
          "The output ends at the 1024-token limit.",
          "This review does not validate every claim in the later text."
        ]
      },
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_122104/gemma26_2048_1234",
      "model": "gemma26",
      "mode": "serial",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 102.14983602142898,
      "metric": "Output token slope after token 32",
      "runtime_tps": 102.40676098637493,
      "ttft_seconds": 0.9243123328778893,
      "gpu_mhz": 1380,
      "peak_load": 6.53271484375,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "37a4ed4fda328209faf30c440eaf9441780b47fafe2bc3600ff935bbea2ef73f",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": null,
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_122104/gemma26_260864_1234",
      "model": "gemma26",
      "mode": "serial",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 11.173547058678103,
      "metric": "Output token slope after token 32",
      "runtime_tps": 11.242669172774159,
      "ttft_seconds": 870.1184073751792,
      "gpu_mhz": 511,
      "peak_load": 12.8564453125,
      "recall": 2,
      "qualified": false,
      "failed_checks": [
        "normal_decode_clock",
        "load"
      ],
      "quality_pass": false,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "e2f26b0becd336e3df256cc346063d938d123bb0c7db58f0d5c58508a52d2636",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "notes": []
      },
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_122104/gemma26_32768_1234",
      "model": "gemma26",
      "mode": "serial",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 71.47543721286803,
      "metric": "Output token slope after token 32",
      "runtime_tps": 71.56874123429077,
      "ttft_seconds": 16.147099709138274,
      "gpu_mhz": 1380,
      "peak_load": 5.36767578125,
      "recall": 6,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": false,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "f45c281855f1e6802e8b6bf21ecb42ef99adcf66a1acae478de3476c5adaa81b",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": null,
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_122104/mimo9_2048_1234",
      "model": "mimo9",
      "mode": "serial",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 98.97610975333399,
      "metric": "Output token slope after token 32",
      "runtime_tps": 99.06963390514048,
      "ttft_seconds": 1.3318278749939054,
      "gpu_mhz": 1380,
      "peak_load": 1.72900390625,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "6c5fcceb83d2cd4d75d7f4d9de9a47c83d3e40a34823e6eb417055ab6a4567c7",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": null,
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "incomplete",
      "error": "KeyboardInterrupt('signal 15')",
      "id": "eight_models_bench/results/run_20261002_122104/mimo9_260864_1234",
      "model": "mimo9",
      "mode": "serial",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": null,
      "output_tokens": null,
      "cached_tokens": null,
      "decode_tps": null,
      "metric": "Output token slope after token 32",
      "runtime_tps": null,
      "ttft_seconds": null,
      "gpu_mhz": null,
      "peak_load": 22.4716796875,
      "recall": null,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": null,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": null,
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": null,
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_122104/mimo9_32768_1234",
      "model": "mimo9",
      "mode": "serial",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 82.51710456088227,
      "metric": "Output token slope after token 32",
      "runtime_tps": 82.71778825314061,
      "ttft_seconds": 23.572585667017847,
      "gpu_mhz": 1380,
      "peak_load": 2.1591796875,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "019fd1f25a694459a88b3ddfb2d905d2f3ee05a9e774400981fd34ed419aeaef",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "notes": []
      },
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "incomplete",
      "error": "RuntimeError('Required cool/quiet gate failed; no timed inference')",
      "id": "eight_models_bench/results/run_20261002_122104/qwen2_2048_1234",
      "model": "qwen2",
      "mode": "serial",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": null,
      "output_tokens": null,
      "cached_tokens": null,
      "decode_tps": null,
      "metric": "Output token slope after token 32",
      "runtime_tps": null,
      "ttft_seconds": null,
      "gpu_mhz": null,
      "peak_load": null,
      "recall": null,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": null,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": null,
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": null,
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_122104/qwen2_2048_1234_mlx-lm_serial_0",
      "model": "qwen2",
      "mode": "serial",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 275.0744830205785,
      "metric": "Output token slope after token 32",
      "runtime_tps": 276.0479600490575,
      "ttft_seconds": 0.3087349580600858,
      "gpu_mhz": 1380,
      "peak_load": 1.70751953125,
      "recall": 8,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": false,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "09da412f50e91400cc79ad20dbc19b72cfe3ca979a4d01157efafdc0fc372be8",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": null,
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_122104/qwen2_260864_1234_mlx-lm_serial_2",
      "model": "qwen2",
      "mode": "serial",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 82.37516970504157,
      "metric": "Output token slope after token 32",
      "runtime_tps": 82.76094248337455,
      "ttft_seconds": 134.66545650013722,
      "gpu_mhz": 1380,
      "peak_load": 2.64697265625,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "36f26b4711b8d531086cd93cbe2ebdec49b79182bd35fc1f4edbdb9022783fc3",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "notes": []
      },
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_122104/qwen2_260864_1235_mlx-lm_serial_3",
      "model": "qwen2",
      "mode": "serial",
      "seed": 1235,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 82.59237889219403,
      "metric": "Output token slope after token 32",
      "runtime_tps": 82.92675428910565,
      "ttft_seconds": 134.23560204193927,
      "gpu_mhz": 1380,
      "peak_load": 3.10791015625,
      "recall": 0,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": false,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "6a36f5ab33ee7c6267a2de440a1a5317b79e9a20dbdbdd1c17437ddc5460dad6",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "notes": []
      },
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_122104/qwen2_260864_1236_mlx-lm_serial_4",
      "model": "qwen2",
      "mode": "serial",
      "seed": 1236,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 82.43007842440414,
      "metric": "Output token slope after token 32",
      "runtime_tps": 82.77643312079879,
      "ttft_seconds": 134.40885775000788,
      "gpu_mhz": 1380,
      "peak_load": 2.68994140625,
      "recall": 8,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": false,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "e78ddffd05f9ff9c51b6a20a71191b4ff864f1e3c2268b667fcea0fe50fc8f33",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "notes": []
      },
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_122104/qwen2_32768_1234_mlx-lm_serial_1",
      "model": "qwen2",
      "mode": "serial",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 215.48362036358375,
      "metric": "Output token slope after token 32",
      "runtime_tps": 216.02883879129956,
      "ttft_seconds": 5.762067708885297,
      "gpu_mhz": 1380,
      "peak_load": 2.06689453125,
      "recall": 0,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": false,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "211e4e515a69783848983b6703b2796d83abd5f61fe1cb13fe9e556962adcd02",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": null,
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_122104/spark17_2048_1234",
      "model": "spark17",
      "mode": "serial",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 196.04597447398277,
      "metric": "Output token slope after token 32",
      "runtime_tps": 196.36280253648866,
      "ttft_seconds": 0.4083410841412842,
      "gpu_mhz": 1380,
      "peak_load": 2.41259765625,
      "recall": 0,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": false,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "849698b821a556ef8e93d440c689e5a292c5333cc3079f3f589d3c34d93643d6",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": null,
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_122104/spark17_260864_1234",
      "model": "spark17",
      "mode": "serial",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 66.58334139697574,
      "metric": "Output token slope after token 32",
      "runtime_tps": 66.84738853843291,
      "ttft_seconds": 158.45800962508656,
      "gpu_mhz": 1380,
      "peak_load": 2.4853515625,
      "recall": 0,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": false,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "1e087ca5544c35a1163641c4901414bbf3679623d09b246d8c61246a7cb4b577",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "notes": []
      },
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_122104/spark17_32768_1234",
      "model": "spark17",
      "mode": "serial",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 158.6257840801755,
      "metric": "Output token slope after token 32",
      "runtime_tps": 158.93043953868087,
      "ttft_seconds": 6.746556208003312,
      "gpu_mhz": 1380,
      "peak_load": 2.4658203125,
      "recall": 0,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": false,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "496c247509e5ccbd4c7933b1bf37c289d79a077aca78116d9d09abde0f67e232",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": null,
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_122104/spark4_2048_1234",
      "model": "spark4",
      "mode": "serial",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 103.05902936845884,
      "metric": "Output token slope after token 32",
      "runtime_tps": 103.18562221439157,
      "ttft_seconds": 0.8421748748514801,
      "gpu_mhz": 1380,
      "peak_load": 1.943359375,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "52f46db46f2d7f4ab7932a494ec57d0c7ea9328aeea7cb5938bb6668d8a367d5",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": null,
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_122104/spark4_260864_1234",
      "model": "spark4",
      "mode": "serial",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 15.609657564932338,
      "metric": "Output token slope after token 32",
      "runtime_tps": 15.846425205805764,
      "ttft_seconds": 987.6044692911673,
      "gpu_mhz": 338,
      "peak_load": 4.01123046875,
      "recall": 8,
      "qualified": false,
      "failed_checks": [
        "normal_decode_clock",
        "decode_clamp",
        "unmodified_execution"
      ],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "fe030f5bfb912b404216d39e1eb821bfa8c68947e651ce4e75d50a1478ba7b9e",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "notes": []
      },
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_122104/spark4_32768_1234",
      "model": "spark4",
      "mode": "serial",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 83.80957741448601,
      "metric": "Output token slope after token 32",
      "runtime_tps": 84.00924811593761,
      "ttft_seconds": 16.497577582951635,
      "gpu_mhz": 1380,
      "peak_load": 2.3017578125,
      "recall": 7,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": false,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "86a0b7f75904d60ad8d06e457ab1eaa938f5b5c2b5a674b40083de2db86575ec",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": null,
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_143653/kat35_2048_1234_mlx-lm_serial_0",
      "model": "kat35",
      "mode": "serial",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 86.82909709741644,
      "metric": "Output token slope after token 32",
      "runtime_tps": 87.0419979783497,
      "ttft_seconds": 0.8578221669886261,
      "gpu_mhz": 1380,
      "peak_load": 2.4697265625,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "8fa5ec63fc319fa0c0b6540b0ac61da4f6ba879fef39fc76fac8c2e971ddd273",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": null,
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "incomplete",
      "error": "KeyboardInterrupt('signal 15')",
      "id": "eight_models_bench/results/run_20261002_143653/kat35_260864_1234_mlx-lm_serial_2",
      "model": "kat35",
      "mode": "serial",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": null,
      "output_tokens": null,
      "cached_tokens": null,
      "decode_tps": null,
      "metric": "Output token slope after token 32",
      "runtime_tps": null,
      "ttft_seconds": null,
      "gpu_mhz": null,
      "peak_load": 25.26708984375,
      "recall": null,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": null,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": null,
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": null,
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_143653/kat35_32768_1234_mlx-lm_serial_1",
      "model": "kat35",
      "mode": "serial",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 76.72082323710272,
      "metric": "Output token slope after token 32",
      "runtime_tps": 76.77316849878159,
      "ttft_seconds": 14.873130082851276,
      "gpu_mhz": 1380,
      "peak_load": 3.103515625,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "10b79c0634e7d68add10f03461828a565b9fef966ef429e9b43739ab9cec5a56",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": null,
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_145828/north30_2048_1234_mlx-lm_serial_0",
      "model": "north30",
      "mode": "serial",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 88.38474025521931,
      "metric": "Output token slope after token 32",
      "runtime_tps": 88.5260087755807,
      "ttft_seconds": 0.9328735829330981,
      "gpu_mhz": 1380,
      "peak_load": 3.03369140625,
      "recall": 0,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": false,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "1bbad2b0374a882abd9330f4be61b4eb49080a48b00778be95bc375e15ed370c",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "notes": []
      },
      "runtime_id": "mlx_lm_north"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "incomplete",
      "error": "KeyboardInterrupt('signal 15')",
      "id": "eight_models_bench/results/run_20261002_145828/north30_260864_1234_mlx-lm_serial_2",
      "model": "north30",
      "mode": "serial",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": null,
      "output_tokens": null,
      "cached_tokens": null,
      "decode_tps": null,
      "metric": "Output token slope after token 32",
      "runtime_tps": null,
      "ttft_seconds": null,
      "gpu_mhz": null,
      "peak_load": null,
      "recall": null,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": null,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": null,
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "notes": []
      },
      "runtime_id": "mlx_lm_north"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_145828/north30_32768_1234_mlx-lm_serial_1",
      "model": "north30",
      "mode": "serial",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 78.80117407256402,
      "metric": "Output token slope after token 32",
      "runtime_tps": 78.94461509113032,
      "ttft_seconds": 20.315452666953206,
      "gpu_mhz": 1380,
      "peak_load": 3.37255859375,
      "recall": 0,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": false,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "415cb8dd28e508718144dcfc45349351ddfdf0be57ee2a7d09d34adcdc892e1d",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "notes": []
      },
      "runtime_id": "mlx_lm_north"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "incomplete",
      "error": "ValueError('Received 225 parameters not in model: \\nmodel.layers.0.attention.k_conv1d.conv.weight,\\nmodel.layers.0.attention.q_conv1d.conv.weight,\\nmodel.layers.0.attention.v_conv1d.conv.weight,\\nmodel.layers.1.attention.k_conv1d.conv.weight,\\nmodel.layers.1.attention.q_conv1d.conv.weight,\\nmodel.layers.1.attention.v_conv1d.conv.weight,\\nmodel.layers.10.attention.k_conv1d.conv.weight,\\nmodel.layers.10.attention.q_conv1d.conv.weight,\\nmodel.layers.10.attention.v_conv1d.conv.weight,\\nmodel.layers.10.mlp.gate.gate_proj.biases,\\nmodel.layers.10.mlp.gate.gate_proj.scales,\\nmodel.layers.10.mlp.gate.gate_proj.weight,\\nmodel.layers.11.mlp.gate.gate_proj.biases,\\nmodel.layers.11.mlp.gate.gate_proj.scales,\\nmodel.layers.11.mlp.gate.gate_proj.weight,\\nmodel.layers.12.attention.k_conv1d.conv.weight,\\nmodel.layers.12.attention.q_conv1d.conv.weight,\\nmodel.layers.12.attention.v_conv1d.conv.weight,\\nmodel.layers.12.mlp.gate.gate_proj.biases,\\nmodel.layers.12.mlp.gate.gate_proj.scales,\\nmodel.layers.12.mlp.gate.gate_proj.weight,\\nmodel.layers.13.attention.k_conv1d.conv.weight,\\nmodel.layers.13.attention.q_conv1d.conv.weight,\\nmodel.layers.13.attention.v_conv1d.conv.weight,\\nmodel.layers.13.mlp.gate.gate_proj.biases,\\nmodel.layers.13.mlp.gate.gate_proj.scales,\\nmodel.layers.13.mlp.gate.gate_proj.weight,\\nmodel.layers.14.attention.k_conv1d.conv.weight,\\nmodel.layers.14.attention.q_conv1d.conv.weight,\\nmodel.layers.14.attention.v_conv1d.conv.weight,\\nmodel.layers.14.mlp.gate.gate_proj.biases,\\nmodel.layers.14.mlp.gate.gate_proj.scales,\\nmodel.layers.14.mlp.gate.gate_proj.weight,\\nmodel.layers.15.attention.k_conv1d.conv.weight,\\nmodel.layers.15.attention.q_conv1d.conv.weight,\\nmodel.layers.15.attention.v_conv1d.conv.weight,\\nmodel.layers.15.mlp.gate.gate_proj.biases,\\nmodel.layers.15.mlp.gate.gate_proj.scales,\\nmodel.layers.15.mlp.gate.gate_proj.weight,\\nmodel.layers.16.attention.k_conv1d.conv.weight,\\nmodel.layers.16.attention.q_conv1d.conv.weight,\\nmodel.layers.16.attention.v_conv1d.conv.weight,\\nmodel.layers.16.mlp.gate.gate_proj.biases,\\nmodel.layers.16.mlp.gate.gate_proj.scales,\\nmodel.layers.16.mlp.gate.gate_proj.weight,\\nmodel.layers.17.mlp.gate.gate_proj.biases,\\nmodel.layers.17.mlp.gate.gate_proj.scales,\\nmodel.layers.17.mlp.gate.gate_proj.weight,\\nmodel.layers.18.attention.k_conv1d.conv.weight,\\nmodel.layers.18.attention.q_conv1d.conv.weight,\\nmodel.layers.18.attention.v_conv1d.conv.weight,\\nmodel.layers.18.mlp.gate.gate_proj.biases,\\nmodel.layers.18.mlp.gate.gate_proj.scales,\\nmodel.layers.18.mlp.gate.gate_proj.weight,\\nmodel.layers.19.attention.k_conv1d.conv.weight,\\nmodel.layers.19.attention.q_conv1d.conv.weight,\\nmodel.layers.19.attention.v_conv1d.conv.weight,\\nmodel.layers.19.mlp.gate.gate_proj.biases,\\nmodel.layers.19.mlp.gate.gate_proj.scales,\\nmodel.layers.19.mlp.gate.gate_proj.weight,\\nmodel.layers.2.attention.k_conv1d.conv.weight,\\nmodel.layers.2.attention.q_conv1d.conv.weight,\\nmodel.layers.2.attention.v_conv1d.conv.weight,\\nmodel.layers.2.mlp.gate.gate_proj.biases,\\nmodel.layers.2.mlp.gate.gate_proj.scales,\\nmodel.layers.2.mlp.gate.gate_proj.weight,\\nmodel.layers.20.attention.k_conv1d.conv.weight,\\nmodel.layers.20.attention.q_conv1d.conv.weight,\\nmodel.layers.20.attention.v_conv1d.conv.weight,\\nmodel.layers.20.mlp.gate.gate_proj.biases,\\nmodel.layers.20.mlp.gate.gate_proj.scales,\\nmodel.layers.20.mlp.gate.gate_proj.weight,\\nmodel.layers.21.attention.k_conv1d.conv.weight,\\nmodel.layers.21.attention.q_conv1d.conv.weight,\\nmodel.layers.21.attention.v_conv1d.conv.weight,\\nmodel.layers.21.mlp.gate.gate_proj.biases,\\nmodel.layers.21.mlp.gate.gate_proj.scales,\\nmodel.layers.21.mlp.gate.gate_proj.weight,\\nmodel.layers.22.attention.k_conv1d.conv.weight,\\nmodel.layers.22.attention.q_conv1d.conv.weight,\\nmodel.layers.22.attention.v_conv1d.conv.weight,\\nmodel.layers.22.mlp.gate.gate_proj.biases,\\nmodel.layers.22.mlp.gate.gate_proj.scales,\\nmodel.layers.22.mlp.gate.gate_proj.weight,\\nmodel.layers.23.mlp.gate.gate_proj.biases,\\nmodel.layers.23.mlp.gate.gate_proj.scales,\\nmodel.layers.23.mlp.gate.gate_proj.weight,\\nmodel.layers.24.attention.k_conv1d.conv.weight,\\nmodel.layers.24.attention.q_conv1d.conv.weight,\\nmodel.layers.24.attention.v_conv1d.conv.weight,\\nmodel.layers.24.mlp.gate.gate_proj.biases,\\nmodel.layers.24.mlp.gate.gate_proj.scales,\\nmodel.layers.24.mlp.gate.gate_proj.weight,\\nmodel.layers.25.attention.k_conv1d.conv.weight,\\nmodel.layers.25.attention.q_conv1d.conv.weight,\\nmodel.layers.25.attention.v_conv1d.conv.weight,\\nmodel.layers.25.mlp.gate.gate_proj.biases,\\nmodel.layers.25.mlp.gate.gate_proj.scales,\\nmodel.layers.25.mlp.gate.gate_proj.weight,\\nmodel.layers.26.attention.k_conv1d.conv.weight,\\nmodel.layers.26.attention.q_conv1d.conv.weight,\\nmodel.layers.26.attention.v_conv1d.conv.weight,\\nmodel.layers.26.mlp.gate.gate_proj.biases,\\nmodel.layers.26.mlp.gate.gate_proj.scales,\\nmodel.layers.26.mlp.gate.gate_proj.weight,\\nmodel.layers.27.attention.k_conv1d.conv.weight,\\nmodel.layers.27.attention.q_conv1d.conv.weight,\\nmodel.layers.27.attention.v_conv1d.conv.weight,\\nmodel.layers.27.mlp.gate.gate_proj.biases,\\nmodel.layers.27.mlp.gate.gate_proj.scales,\\nmodel.layers.27.mlp.gate.gate_proj.weight,\\nmodel.layers.28.attention.k_conv1d.conv.weight,\\nmodel.layers.28.attention.q_conv1d.conv.weight,\\nmodel.layers.28.attention.v_conv1d.conv.weight,\\nmodel.layers.28.mlp.gate.gate_proj.biases,\\nmodel.layers.28.mlp.gate.gate_proj.scales,\\nmodel.layers.28.mlp.gate.gate_proj.weight,\\nmodel.layers.29.mlp.gate.gate_proj.biases,\\nmodel.layers.29.mlp.gate.gate_proj.scales,\\nmodel.layers.29.mlp.gate.gate_proj.weight,\\nmodel.layers.3.attention.k_conv1d.conv.weight,\\nmodel.layers.3.attention.q_conv1d.conv.weight,\\nmodel.layers.3.attention.v_conv1d.conv.weight,\\nmodel.layers.3.mlp.gate.gate_proj.biases,\\nmodel.layers.3.mlp.gate.gate_proj.scales,\\nmodel.layers.3.mlp.gate.gate_proj.weight,\\nmodel.layers.30.attention.k_conv1d.conv.weight,\\nmodel.layers.30.attention.q_conv1d.conv.weight,\\nmodel.layers.30.attention.v_conv1d.conv.weight,\\nmodel.layers.30.mlp.gate.gate_proj.biases,\\nmodel.layers.30.mlp.gate.gate_proj.scales,\\nmodel.layers.30.mlp.gate.gate_proj.weight,\\nmodel.layers.31.attention.k_conv1d.conv.weight,\\nmodel.layers.31.attention.q_conv1d.conv.weight,\\nmodel.layers.31.attention.v_conv1d.conv.weight,\\nmodel.layers.31.mlp.gate.gate_proj.biases,\\nmodel.layers.31.mlp.gate.gate_proj.scales,\\nmodel.layers.31.mlp.gate.gate_proj.weight,\\nmodel.layers.32.attention.k_conv1d.conv.weight,\\nmodel.layers.32.attention.q_conv1d.conv.weight,\\nmodel.layers.32.attention.v_conv1d.conv.weight,\\nmodel.layers.32.mlp.gate.gate_proj.biases,\\nmodel.layers.32.mlp.gate.gate_proj.scales,\\nmodel.layers.32.mlp.gate.gate_proj.weight,\\nmodel.layers.33.attention.k_conv1d.conv.weight,\\nmodel.layers.33.attention.q_conv1d.conv.weight,\\nmodel.layers.33.attention.v_conv1d.conv.weight,\\nmodel.layers.33.mlp.gate.gate_proj.biases,\\nmodel.layers.33.mlp.gate.gate_proj.scales,\\nmodel.layers.33.mlp.gate.gate_proj.weight,\\nmodel.layers.34.attention.k_conv1d.conv.weight,\\nmodel.layers.34.attention.q_conv1d.conv.weight,\\nmodel.layers.34.attention.v_conv1d.conv.weight,\\nmodel.layers.34.mlp.gate.gate_proj.biases,\\nmodel.layers.34.mlp.gate.gate_proj.scales,\\nmodel.layers.34.mlp.gate.gate_proj.weight,\\nmodel.layers.35.mlp.gate.gate_proj.biases,\\nmodel.layers.35.mlp.gate.gate_proj.scales,\\nmodel.layers.35.mlp.gate.gate_proj.weight,\\nmodel.layers.36.attention.k_conv1d.conv.weight,\\nmodel.layers.36.attention.q_conv1d.conv.weight,\\nmodel.layers.36.attention.v_conv1d.conv.weight,\\nmodel.layers.36.mlp.gate.gate_proj.biases,\\nmodel.layers.36.mlp.gate.gate_proj.scales,\\nmodel.layers.36.mlp.gate.gate_proj.weight,\\nmodel.layers.37.attention.k_conv1d.conv.weight,\\nmodel.layers.37.attention.q_conv1d.conv.weight,\\nmodel.layers.37.attention.v_conv1d.conv.weight,\\nmodel.layers.37.mlp.gate.gate_proj.biases,\\nmodel.layers.37.mlp.gate.gate_proj.scales,\\nmodel.layers.37.mlp.gate.gate_proj.weight,\\nmodel.layers.38.attention.k_conv1d.conv.weight,\\nmodel.layers.38.attention.q_conv1d.conv.weight,\\nmodel.layers.38.attention.v_conv1d.conv.weight,\\nmodel.layers.38.mlp.gate.gate_proj.biases,\\nmodel.layers.38.mlp.gate.gate_proj.scales,\\nmodel.layers.38.mlp.gate.gate_proj.weight,\\nmodel.layers.39.attention.k_conv1d.conv.weight,\\nmodel.layers.39.attention.q_conv1d.conv.weight,\\nmodel.layers.39.attention.v_conv1d.conv.weight,\\nmodel.layers.39.mlp.gate.gate_proj.biases,\\nmodel.layers.39.mlp.gate.gate_proj.scales,\\nmodel.layers.39.mlp.gate.gate_proj.weight,\\nmodel.layers.4.attention.k_conv1d.conv.weight,\\nmodel.layers.4.attention.q_conv1d.conv.weight,\\nmodel.layers.4.attention.v_conv1d.conv.weight,\\nmodel.layers.4.mlp.gate.gate_proj.biases,\\nmodel.layers.4.mlp.gate.gate_proj.scales,\\nmodel.layers.4.mlp.gate.gate_proj.weight,\\nmodel.layers.40.attention.k_conv1d.conv.weight,\\nmodel.layers.40.attention.q_conv1d.conv.weight,\\nmodel.layers.40.attention.v_conv1d.conv.weight,\\nmodel.layers.40.mlp.gate.gate_proj.biases,\\nmodel.layers.40.mlp.gate.gate_proj.scales,\\nmodel.layers.40.mlp.gate.gate_proj.weight,\\nmodel.layers.41.mlp.gate.gate_proj.biases,\\nmodel.layers.41.mlp.gate.gate_proj.scales,\\nmodel.layers.41.mlp.gate.gate_proj.weight,\\nmodel.layers.5.mlp.gate.gate_proj.biases,\\nmodel.layers.5.mlp.gate.gate_proj.scales,\\nmodel.layers.5.mlp.gate.gate_proj.weight,\\nmodel.layers.6.attention.k_conv1d.conv.weight,\\nmodel.layers.6.attention.q_conv1d.conv.weight,\\nmodel.layers.6.attention.v_conv1d.conv.weight,\\nmodel.layers.6.mlp.gate.gate_proj.biases,\\nmodel.layers.6.mlp.gate.gate_proj.scales,\\nmodel.layers.6.mlp.gate.gate_proj.weight,\\nmodel.layers.7.attention.k_conv1d.conv.weight,\\nmodel.layers.7.attention.q_conv1d.conv.weight,\\nmodel.layers.7.attention.v_conv1d.conv.weight,\\nmodel.layers.7.mlp.gate.gate_proj.biases,\\nmodel.layers.7.mlp.gate.gate_proj.scales,\\nmodel.layers.7.mlp.gate.gate_proj.weight,\\nmodel.layers.8.attention.k_conv1d.conv.weight,\\nmodel.layers.8.attention.q_conv1d.conv.weight,\\nmodel.layers.8.attention.v_conv1d.conv.weight,\\nmodel.layers.8.mlp.gate.gate_proj.biases,\\nmodel.layers.8.mlp.gate.gate_proj.scales,\\nmodel.layers.8.mlp.gate.gate_proj.weight,\\nmodel.layers.9.attention.k_conv1d.conv.weight,\\nmodel.layers.9.attention.q_conv1d.conv.weight,\\nmodel.layers.9.attention.v_conv1d.conv.weight,\\nmodel.layers.9.mlp.gate.gate_proj.biases,\\nmodel.layers.9.mlp.gate.gate_proj.scales,\\nmodel.layers.9.mlp.gate.gate_proj.weight.')",
      "id": "eight_models_bench/results/run_20261002_150547/ling124_2048_1234_mlx-lm_serial_3",
      "model": "ling124",
      "mode": "serial",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": null,
      "output_tokens": null,
      "cached_tokens": null,
      "decode_tps": null,
      "metric": "Output token slope after token 32",
      "runtime_tps": null,
      "ttft_seconds": null,
      "gpu_mhz": null,
      "peak_load": null,
      "recall": null,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": null,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": null,
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": null,
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_150547/north30_2048_1234_mlx-lm_serial_0",
      "model": "north30",
      "mode": "serial",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 88.49229597823052,
      "metric": "Output token slope after token 32",
      "runtime_tps": 88.52222412785842,
      "ttft_seconds": 0.9313536249101162,
      "gpu_mhz": 1380,
      "peak_load": 1.97021484375,
      "recall": 7,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": false,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "1c4065ecea2fb2c76b2419b9905a4b878bfba0c58ea968cf0a73e99849739827",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": null,
      "runtime_id": "mlx_lm_north"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "incomplete",
      "error": "KeyboardInterrupt('signal 15')",
      "id": "eight_models_bench/results/run_20261002_150547/north30_260864_1234_mlx-lm_serial_2",
      "model": "north30",
      "mode": "serial",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": null,
      "output_tokens": null,
      "cached_tokens": null,
      "decode_tps": null,
      "metric": "Output token slope after token 32",
      "runtime_tps": null,
      "ttft_seconds": null,
      "gpu_mhz": null,
      "peak_load": 27.99951171875,
      "recall": null,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": null,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": null,
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": null,
      "runtime_id": "mlx_lm_north"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_150547/north30_32768_1234_mlx-lm_serial_1",
      "model": "north30",
      "mode": "serial",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 78.14366061809876,
      "metric": "Output token slope after token 32",
      "runtime_tps": 78.30870754991905,
      "ttft_seconds": 20.31380237499252,
      "gpu_mhz": 1380,
      "peak_load": 2.37548828125,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "a1b32721f795f3198b0917cea3ad21eb1b33e5aabccc0918d4e595227f9a210b",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": null,
      "runtime_id": "mlx_lm_north"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_153149/ling124_2048_1234_mlx-lm_serial_0",
      "model": "ling124",
      "mode": "serial",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 64.4600221243539,
      "metric": "Output token slope after token 32",
      "runtime_tps": 64.48037273254289,
      "ttft_seconds": 1.8869427498430014,
      "gpu_mhz": 1380,
      "peak_load": 4.87939453125,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "6bdc944d2044b65cffdcb2768e2c7ec52b3ef3844714ca52feb76f85e2c1e8e5",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": null,
      "adapter_sha256": "8bac095afcbce4a80f200d9a469bd7febac058de9b7f70ea4f3c006faeedff78",
      "runtime_id": "mlx_lm_ling"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "incomplete",
      "error": "KeyboardInterrupt('signal 15')",
      "id": "eight_models_bench/results/run_20261002_153149/ling124_260864_1234_mlx-lm_serial_2",
      "model": "ling124",
      "mode": "serial",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": null,
      "output_tokens": null,
      "cached_tokens": null,
      "decode_tps": null,
      "metric": "Output token slope after token 32",
      "runtime_tps": null,
      "ttft_seconds": null,
      "gpu_mhz": null,
      "peak_load": 26.42724609375,
      "recall": null,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": null,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": null,
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": null,
      "adapter_sha256": "8bac095afcbce4a80f200d9a469bd7febac058de9b7f70ea4f3c006faeedff78",
      "runtime_id": "mlx_lm_ling"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_153149/ling124_32768_1234_mlx-lm_serial_1",
      "model": "ling124",
      "mode": "serial",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 929,
      "cached_tokens": 0,
      "decode_tps": 59.010190373089976,
      "metric": "Output token slope after token 32",
      "runtime_tps": 59.08595835076994,
      "ttft_seconds": 25.146480708150193,
      "gpu_mhz": 1380,
      "peak_load": 4.3505859375,
      "recall": 8,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "7d29d2bbcd19c640ace883f6333e1eb161f2ac625da83a28c11c53634e73271d",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": null,
      "adapter_sha256": "8bac095afcbce4a80f200d9a469bd7febac058de9b7f70ea4f3c006faeedff78",
      "runtime_id": "mlx_lm_ling"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "incomplete",
      "error": "KeyboardInterrupt('signal 15')",
      "id": "eight_models_bench/results/run_20261002_153149/mimo9_260864_1234_mlx-lm_serial_3",
      "model": "mimo9",
      "mode": "serial_diagnostic_load",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": null,
      "output_tokens": null,
      "cached_tokens": null,
      "decode_tps": null,
      "metric": "Output token slope after token 32",
      "runtime_tps": null,
      "ttft_seconds": null,
      "gpu_mhz": null,
      "peak_load": 4.14453125,
      "recall": null,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": null,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": null,
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": null,
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_160205/kat35_260864_1234_mlx-lm_serial_10",
      "model": "kat35",
      "mode": "serial_prefill_pause2000ms",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 13.266754232792485,
      "metric": "Output token slope after token 32",
      "runtime_tps": 13.685817792670521,
      "ttft_seconds": 1084.4261728751007,
      "gpu_mhz": 338,
      "peak_load": 8.23046875,
      "recall": 8,
      "qualified": false,
      "failed_checks": [
        "normal_decode_clock",
        "decode_clamp",
        "load"
      ],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 511.418514584424,
      "token_sha256": "d943c66e5579b07ded2345f2b84c3a798415ceec5af4a98c4a6939b3c4064ec9",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "notes": []
      },
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_160205/mimo9_2048_1234_mlx-lm_serial_0",
      "model": "mimo9",
      "mode": "serial_prefill_pause2000ms",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 98.85730077245455,
      "metric": "Output token slope after token 32",
      "runtime_tps": 98.9623808871321,
      "ttft_seconds": 5.4173535839654505,
      "gpu_mhz": 1380,
      "peak_load": 2.0107421875,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 4.014204375213012,
      "token_sha256": "6c5fcceb83d2cd4d75d7f4d9de9a47c83d3e40a34823e6eb417055ab6a4567c7",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": null,
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_160205/mimo9_260864_1234_mlx-lm_serial_9",
      "model": "mimo9",
      "mode": "serial_prefill_pause2000ms",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 15.807108310310591,
      "metric": "Output token slope after token 32",
      "runtime_tps": 16.308729386313146,
      "ttft_seconds": 1448.8684364999644,
      "gpu_mhz": 338,
      "peak_load": 8.26416015625,
      "recall": 8,
      "qualified": false,
      "failed_checks": [
        "normal_decode_clock",
        "decode_clamp",
        "load"
      ],
      "quality_pass": false,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 511.16780873783864,
      "token_sha256": "446052b85e70811ee2ce7d8b8d1eac5ecc7467b18d6ee1fc906da141c017b56a",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "notes": []
      },
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_160205/qwen2_2048_1234_mlx-lm_serial_1",
      "model": "qwen2",
      "mode": "serial_kv8_after_prefill",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 265.86132054695145,
      "metric": "Output token slope after token 32",
      "runtime_tps": 266.23036204534304,
      "ttft_seconds": 0.3409620001912117,
      "gpu_mhz": 1380,
      "peak_load": 2.0625,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": 8,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "1d811c65dc8117214959a48707d405da609b1f131728698a5919ef5da0fef5a8",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": null,
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_160205/qwen2_2048_1234_mlx-lm_serial_6",
      "model": "qwen2",
      "mode": "serial_qwen_recommended",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 253.56652339929337,
      "metric": "Output token slope after token 32",
      "runtime_tps": 254.01674835226095,
      "ttft_seconds": 0.3102220417931676,
      "gpu_mhz": 1380,
      "peak_load": 2.2705078125,
      "recall": 0,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": false,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "70997a8c912e1a4a01d1a2a25b4582414550f74bdf1fc2d36f0fa2ef83fa3c97",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "notes": []
      },
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_160205/qwen2_260864_1234_mlx-lm_serial_3",
      "model": "qwen2",
      "mode": "serial_kv8_after_prefill",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 87.76879038762831,
      "metric": "Output token slope after token 32",
      "runtime_tps": 88.01081862655354,
      "ttft_seconds": 134.6763551668264,
      "gpu_mhz": 1380,
      "peak_load": 2.427734375,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": 8,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "f9697e21fd25128ebfc4ed8f8e057eda9daf3c079be502db599339cb3f893c7b",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "notes": []
      },
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_160205/qwen2_260864_1234_mlx-lm_serial_8",
      "model": "qwen2",
      "mode": "serial_qwen_recommended",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 80.37162222609965,
      "metric": "Output token slope after token 32",
      "runtime_tps": 80.65752158504752,
      "ttft_seconds": 134.55813529202715,
      "gpu_mhz": 1380,
      "peak_load": 1.919921875,
      "recall": 0,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": false,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "5f1a1ece6e870e0bfb2de40449b436ace4e897f5ebb418ba6fda477b7ea22d7f",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "notes": []
      },
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_160205/qwen2_260864_1235_mlx-lm_serial_4",
      "model": "qwen2",
      "mode": "serial_kv8_after_prefill",
      "seed": 1235,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 87.13979196461283,
      "metric": "Output token slope after token 32",
      "runtime_tps": 87.31575540291995,
      "ttft_seconds": 135.07107970886864,
      "gpu_mhz": 1373,
      "peak_load": 1.69140625,
      "recall": 0,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": false,
      "exact_pair": null,
      "kv_bits": 8,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "793ef2a2c2858cd5d1b16ef5a50da995b8f48e3f6129b451b475c93ecfb89144",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "notes": []
      },
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_160205/qwen2_260864_1236_mlx-lm_serial_5",
      "model": "qwen2",
      "mode": "serial_kv8_after_prefill",
      "seed": 1236,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 87.61887062300919,
      "metric": "Output token slope after token 32",
      "runtime_tps": 87.88393632880673,
      "ttft_seconds": 134.55814587511122,
      "gpu_mhz": 1380,
      "peak_load": 2.71435546875,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": 8,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "c21ae0a1b5142f075d3371055fc4ccff78acf472befe256f54494876617e11c5",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "notes": []
      },
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_160205/qwen2_32768_1234_mlx-lm_serial_2",
      "model": "qwen2",
      "mode": "serial_kv8_after_prefill",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 215.88881614180053,
      "metric": "Output token slope after token 32",
      "runtime_tps": 216.31658121883186,
      "ttft_seconds": 5.767181457951665,
      "gpu_mhz": 1380,
      "peak_load": 1.79833984375,
      "recall": 0,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": false,
      "exact_pair": null,
      "kv_bits": 8,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "33b045b7d0a60f93e76b078a647a20852d521a9192847bb1ae9b77e8369e4f82",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": null,
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_160205/qwen2_32768_1234_mlx-lm_serial_7",
      "model": "qwen2",
      "mode": "serial_qwen_recommended",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 200.95283048083655,
      "metric": "Output token slope after token 32",
      "runtime_tps": 201.41529816967014,
      "ttft_seconds": 5.7658616250846535,
      "gpu_mhz": 1380,
      "peak_load": 2.06982421875,
      "recall": 0,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": false,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "4c88dfcfc94caa332232db8fa4369eaa352593e79f811dd9595ded06cd5eec09",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "notes": []
      },
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_173233/gemma26_2048_1234_gemma-vlm_mtp_8",
      "model": "gemma26",
      "mode": "gemma-vlm_mtp",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 117.6047946284242,
      "metric": "Output token slope after token 32",
      "runtime_tps": 119.28294566047818,
      "ttft_seconds": 0.8084698328748345,
      "gpu_mhz": 1380,
      "peak_load": 3.86279296875,
      "recall": 8,
      "qualified": false,
      "failed_checks": [
        "matched_serial_exactness",
        "valid_harness"
      ],
      "quality_pass": true,
      "exact_pair": false,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "729d57864a5e201b6f2e974b82e1ca675574872b65f568edf3d54920d1cb8e7b",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": null,
      "runtime_id": "mlx_vlm_gemma"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_173233/gemma26_2048_1234_gemma-vlm_serial_7",
      "model": "gemma26",
      "mode": "gemma-vlm_serial",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 99.1373589591311,
      "metric": "Output token slope after token 32",
      "runtime_tps": 99.23338892428025,
      "ttft_seconds": 0.8164700828492641,
      "gpu_mhz": 1380,
      "peak_load": 4.13427734375,
      "recall": 8,
      "qualified": false,
      "failed_checks": [
        "valid_harness"
      ],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "37a4ed4fda328209faf30c440eaf9441780b47fafe2bc3600ff935bbea2ef73f",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": null,
      "runtime_id": "mlx_vlm_gemma"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_173233/ling124_260864_1234_mlx-lm_serial_6",
      "model": "ling124",
      "mode": "serial_prefill_pause2000ms",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 22.08929035315399,
      "metric": "Output token slope after token 32",
      "runtime_tps": 22.84086689132715,
      "ttft_seconds": 1492.3430594999809,
      "gpu_mhz": 649,
      "peak_load": 6.61376953125,
      "recall": 8,
      "qualified": false,
      "failed_checks": [
        "normal_decode_clock"
      ],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 511.1994058333803,
      "token_sha256": "e013f32096799a06675d29d0973366d480300a281923edde3d4ed25cc19394c0",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "notes": []
      },
      "adapter_sha256": "8bac095afcbce4a80f200d9a469bd7febac058de9b7f70ea4f3c006faeedff78",
      "runtime_id": "mlx_lm_ling"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_173233/north30_260864_1234_mlx-lm_serial_5",
      "model": "north30",
      "mode": "serial_prefill_pause2000ms",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 75,
      "cached_tokens": 0,
      "decode_tps": 42.49555587739015,
      "metric": "Output token slope after token 32",
      "runtime_tps": 43.13961593545853,
      "ttft_seconds": 1056.113874041941,
      "gpu_mhz": 1378,
      "peak_load": 5.37451171875,
      "recall": 4,
      "qualified": false,
      "failed_checks": [
        "decode_samples"
      ],
      "quality_pass": false,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 511.2922780360095,
      "token_sha256": "0776157d493048c6334571e8b9118850e85625fb3ba177df6f27d9885018983a",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "notes": []
      },
      "runtime_id": "mlx_lm_north"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_173233/qwen2_2048_1234_mlx-lm_serial_0",
      "model": "qwen2",
      "mode": "serial_kv4_after_prefill",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 263.9656001080089,
      "metric": "Output token slope after token 32",
      "runtime_tps": 264.85787360997625,
      "ttft_seconds": 0.330167958047241,
      "gpu_mhz": 1380,
      "peak_load": 3.69921875,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": 4,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "76f4216d3fe068373a5ce9607d8c6fe72c790d3244bc0ff3f8da10d42f42b041",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "notes": []
      },
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_173233/qwen2_260864_1234_mlx-lm_serial_2",
      "model": "qwen2",
      "mode": "serial_kv4_after_prefill",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 97.74422210794675,
      "metric": "Output token slope after token 32",
      "runtime_tps": 97.87275736329624,
      "ttft_seconds": 136.47050416679122,
      "gpu_mhz": 1380,
      "peak_load": 4.87744140625,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": 4,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "d219ad28f087273614f3ba4ce92dcee26d416ed423059d5328f1cfe81dc46f5c",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "notes": []
      },
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_173233/qwen2_260864_1235_mlx-lm_serial_3",
      "model": "qwen2",
      "mode": "serial_kv4_after_prefill",
      "seed": 1235,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 97.8040584084113,
      "metric": "Output token slope after token 32",
      "runtime_tps": 97.9479017845984,
      "ttft_seconds": 137.02535220887512,
      "gpu_mhz": 1380,
      "peak_load": 3.77294921875,
      "recall": 0,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": false,
      "exact_pair": null,
      "kv_bits": 4,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "0df1a065e3ec7b8955c9ab3ea2ba11a2a512ced08dab0351c55c8e4007571095",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "notes": []
      },
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_173233/qwen2_260864_1236_mlx-lm_serial_4",
      "model": "qwen2",
      "mode": "serial_kv4_after_prefill",
      "seed": 1236,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 97.85779606748144,
      "metric": "Output token slope after token 32",
      "runtime_tps": 97.99245259519425,
      "ttft_seconds": 136.4322057501413,
      "gpu_mhz": 1380,
      "peak_load": 3.8408203125,
      "recall": 0,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": false,
      "exact_pair": null,
      "kv_bits": 4,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "85396bad56fb18d5fcf376ed53245fc758b4575278f1fa3428d32a902ae6d69f",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "notes": []
      },
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_173233/qwen2_32768_1234_mlx-lm_serial_1",
      "model": "qwen2",
      "mode": "serial_kv4_after_prefill",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 222.0397732487465,
      "metric": "Output token slope after token 32",
      "runtime_tps": 222.47935732388032,
      "ttft_seconds": 5.774283292004839,
      "gpu_mhz": 1380,
      "peak_load": 3.91748046875,
      "recall": 0,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": false,
      "exact_pair": null,
      "kv_bits": 4,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "eabeb5a1e42579f71cdf1b876056acdbd93326b4d2d18b14a27c0f8411d08ed7",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "notes": []
      },
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "incomplete",
      "error": "RuntimeError('Fresh GEMM failed: 17.609271709172713')",
      "id": "eight_models_bench/results/run_20261002_191124/gemma26_2048_1234_gemma-vlm_mtp_1",
      "model": "gemma26",
      "mode": "gemma-vlm_mtp",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": null,
      "output_tokens": null,
      "cached_tokens": null,
      "decode_tps": null,
      "metric": "Output token slope after token 32",
      "runtime_tps": null,
      "ttft_seconds": null,
      "gpu_mhz": null,
      "peak_load": null,
      "recall": null,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": null,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": null,
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": null,
      "runtime_id": "mlx_vlm_gemma"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_191124/gemma26_2048_1234_gemma-vlm_serial_0",
      "model": "gemma26",
      "mode": "gemma-vlm_serial",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 97.023772256582,
      "metric": "Output token slope after token 32",
      "runtime_tps": 96.32372039323822,
      "ttft_seconds": 0.8157739581074566,
      "gpu_mhz": 1380,
      "peak_load": 2.78466796875,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "37a4ed4fda328209faf30c440eaf9441780b47fafe2bc3600ff935bbea2ef73f",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "notes": []
      },
      "runtime_id": "mlx_vlm_gemma"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_191509/gemma26_2048_1234_gemma-vlm_mtp_1",
      "model": "gemma26",
      "mode": "gemma-vlm_mtp",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 116.41290950723713,
      "metric": "Output token slope after token 32",
      "runtime_tps": 118.42403709621364,
      "ttft_seconds": 0.808350375154987,
      "gpu_mhz": 1380,
      "peak_load": 2.6318359375,
      "recall": 8,
      "qualified": false,
      "failed_checks": [
        "matched_serial_exactness"
      ],
      "quality_pass": true,
      "exact_pair": false,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "729d57864a5e201b6f2e974b82e1ca675574872b65f568edf3d54920d1cb8e7b",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "notes": []
      },
      "runtime_id": "mlx_vlm_gemma"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_191509/gemma26_2048_1234_gemma-vlm_serial_0",
      "model": "gemma26",
      "mode": "gemma-vlm_serial",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 96.6840252742502,
      "metric": "Output token slope after token 32",
      "runtime_tps": 96.75448560722353,
      "ttft_seconds": 0.8176984579768032,
      "gpu_mhz": 1380,
      "peak_load": 2.271484375,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "37a4ed4fda328209faf30c440eaf9441780b47fafe2bc3600ff935bbea2ef73f",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "notes": []
      },
      "runtime_id": "mlx_vlm_gemma"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_191509/mimo9_mtplx_2048_1234_mtplx_mtp_3",
      "model": "mimo9_mtplx",
      "mode": "mtplx_mtp",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 117.046370251602,
      "metric": "Output token slope after token 32",
      "runtime_tps": 116.70407406736501,
      "ttft_seconds": 1.5409953340422362,
      "gpu_mhz": 1379,
      "peak_load": 3.771484375,
      "recall": 8,
      "qualified": false,
      "failed_checks": [
        "matched_serial_exactness"
      ],
      "quality_pass": true,
      "exact_pair": false,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "3e606234be33dc5a5ebbd567d3030361455079c21ee8041bbb95285c8bbd3890",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "notes": []
      },
      "runtime_id": "mtplx2120_mimo"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_191509/mimo9_mtplx_2048_1234_mtplx_serial_2",
      "model": "mimo9_mtplx",
      "mode": "mtplx_serial",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 70.3044965148431,
      "metric": "Output token slope after token 32",
      "runtime_tps": 70.29549904011868,
      "ttft_seconds": 1.461706459056586,
      "gpu_mhz": 1380,
      "peak_load": 3.44970703125,
      "recall": 8,
      "qualified": false,
      "failed_checks": [
        "no_external_probe"
      ],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": 16,
      "kv_group_size": 64,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "4cf6ee5570c0c8245db58be45ab5ad4822622a98dc3531a0ea1ff5ed1d9d0ff0",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "notes": []
      },
      "runtime_id": "mtplx2120_mimo"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_191509/qwen2_2048_1234_mlx-lm_serial_4",
      "model": "qwen2",
      "mode": "serial_kv4_g128_after_prefill",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 262.9014537533153,
      "metric": "Output token slope after token 32",
      "runtime_tps": 263.7434216688541,
      "ttft_seconds": 0.33387741702608764,
      "gpu_mhz": 1380,
      "peak_load": 3.83642578125,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": 4,
      "kv_group_size": 128,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "cd67f8f649ae9668dddd703c48a0f897cbd66b078a7525df6413a981e9bf0526",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "notes": []
      },
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_191509/qwen2_260864_1234_mlx-lm_serial_6",
      "model": "qwen2",
      "mode": "serial_kv4_g128_after_prefill",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 98.31397342102457,
      "metric": "Output token slope after token 32",
      "runtime_tps": 98.41987089593361,
      "ttft_seconds": 136.6003080001101,
      "gpu_mhz": 1380,
      "peak_load": 5.73828125,
      "recall": 0,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": false,
      "exact_pair": null,
      "kv_bits": 4,
      "kv_group_size": 128,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "e2f98f7238363d80e9fc19a0f973a3536dd9dacfbde56a73c5fd88194cc60085",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "notes": []
      },
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_191509/qwen2_260864_1235_mlx-lm_serial_7",
      "model": "qwen2",
      "mode": "serial_kv4_g128_after_prefill",
      "seed": 1235,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 98.4785245109778,
      "metric": "Output token slope after token 32",
      "runtime_tps": 98.21226149225882,
      "ttft_seconds": 135.4709887090139,
      "gpu_mhz": 1380,
      "peak_load": 7.1552734375,
      "recall": 0,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": false,
      "exact_pair": null,
      "kv_bits": 4,
      "kv_group_size": 128,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "bfb748f82eea19fbb92f06fde06cfd55e1a629d0dfe425450a4e66bced58b3e3",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "notes": []
      },
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_191509/qwen2_260864_1236_mlx-lm_serial_8",
      "model": "qwen2",
      "mode": "serial_kv4_g128_after_prefill",
      "seed": 1236,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 98.79615720994302,
      "metric": "Output token slope after token 32",
      "runtime_tps": 98.8934484662115,
      "ttft_seconds": 134.34373149997555,
      "gpu_mhz": 1380,
      "peak_load": 3.96533203125,
      "recall": 0,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": false,
      "exact_pair": null,
      "kv_bits": 4,
      "kv_group_size": 128,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "f3f1b74a22889e67fcabc6da28ad0427805c4274867aeeefc0e1e0d989c90ecc",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "notes": []
      },
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "eight",
      "source": "eight_models_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "eight_models_bench/results/run_20261002_191509/qwen2_32768_1234_mlx-lm_serial_5",
      "model": "qwen2",
      "mode": "serial_kv4_g128_after_prefill",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 217.09147030313022,
      "metric": "Output token slope after token 32",
      "runtime_tps": 216.207500532176,
      "ttft_seconds": 5.857985957991332,
      "gpu_mhz": 1380,
      "peak_load": 4.08740234375,
      "recall": 0,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": false,
      "exact_pair": null,
      "kv_bits": 4,
      "kv_group_size": 128,
      "cache_format": null,
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "0f17a7409c4b9ce87951865393bc64d5aa5b675e17ab9c75ad4dde63c87badac",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "notes": []
      },
      "runtime_id": "mlx_lm_eight"
    },
    {
      "campaign": "glm065",
      "source": "glm53_tensorfold065_bench/all-results.json",
      "status": "incomplete",
      "error": "TimeoutError('TensorFold startup exceeded 1440 seconds')",
      "id": "glm53_tensorfold065_bench/results/run_20261005_142827/glm_4bit_2048_1234_tensorfold_serial_0",
      "model": "glm_4bit",
      "mode": "tensorfold_serial",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": null,
      "output_tokens": null,
      "cached_tokens": null,
      "decode_tps": null,
      "metric": "Output token slope after token 32",
      "runtime_tps": null,
      "ttft_seconds": null,
      "gpu_mhz": null,
      "peak_load": null,
      "recall": null,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": null,
      "exact_pair": null,
      "kv_bits": null,
      "kv_group_size": null,
      "cache_format": "BF16 latent and indexer arrays; FP32 KDA state",
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": null,
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": null,
      "setup_source": "glm53_tensorfold065_bench/results/run_20261005_142827/glm_4bit_2048_1234_tensorfold_serial_0/result.json",
      "host_weight_preread_seconds": null,
      "host_weight_preread_bytes": null,
      "host_weight_preread_enabled": false,
      "startup_timeout_seconds": null,
      "setup_excluded_from_ttft": true,
      "eager_weight_load": false,
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "glm065",
      "source": "glm53_tensorfold065_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "glm53_tensorfold065_bench/results/run_20261005_145619/glm_4bit_2048_1234_tensorfold_mtp_1",
      "model": "glm_4bit",
      "mode": "tensorfold_mtp",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 52.07597837899606,
      "metric": "Output token slope after token 32",
      "runtime_tps": 52.05205980972866,
      "ttft_seconds": 5.028661416145042,
      "gpu_mhz": 1380,
      "peak_load": 2.9873046875,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": true,
      "kv_bits": null,
      "kv_group_size": null,
      "cache_format": "BF16 latent and indexer arrays; FP32 KDA state",
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "b9ba2332c6c04964f514a0528054f49084ee49fc232bae6d28f8564e8cadc8d3",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "The output bytes equal the manually reviewed serial output.",
          "All 1024 tokens match the serial control.",
          "All eight answer records match the input.",
          "The same guide limitation applies: one quoted filler sentence is absent from the input.",
          "This review does not validate every guide claim."
        ]
      },
      "setup_source": "glm53_tensorfold065_bench/results/run_20261005_145619/glm_4bit_2048_1234_tensorfold_mtp_1/result.json",
      "host_weight_preread_seconds": 13.602700792020187,
      "host_weight_preread_bytes": 181709451790,
      "host_weight_preread_enabled": true,
      "startup_timeout_seconds": 3600,
      "setup_excluded_from_ttft": true,
      "eager_weight_load": false,
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "glm065",
      "source": "glm53_tensorfold065_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "glm53_tensorfold065_bench/results/run_20261005_145619/glm_4bit_2048_1234_tensorfold_serial_0",
      "model": "glm_4bit",
      "mode": "tensorfold_serial",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 42.692248462523565,
      "metric": "Output token slope after token 32",
      "runtime_tps": 42.68356303478662,
      "ttft_seconds": 5.144392250105739,
      "gpu_mhz": 1380,
      "peak_load": 3.5224609375,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": null,
      "kv_group_size": null,
      "cache_format": "BF16 latent and indexer arrays; FP32 KDA state",
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "b9ba2332c6c04964f514a0528054f49084ee49fc232bae6d28f8564e8cadc8d3",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "All eight answer records match the input.",
          "The output contains no visible repetition loop.",
          "The guide ends at the 1024-token output limit.",
          "One quoted filler sentence does not occur in the input.",
          "The Nakhodka context description agrees with the input.",
          "This review does not validate every guide claim."
        ]
      },
      "setup_source": "glm53_tensorfold065_bench/results/run_20261005_145619/glm_4bit_2048_1234_tensorfold_serial_0/result.json",
      "host_weight_preread_seconds": 1557.5312059579883,
      "host_weight_preread_bytes": 181709451790,
      "host_weight_preread_enabled": true,
      "startup_timeout_seconds": 3600,
      "setup_excluded_from_ttft": true,
      "eager_weight_load": false,
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "glm065",
      "source": "glm53_tensorfold065_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "glm53_tensorfold065_bench/results/run_20261005_145619/glm_4bit_260864_1234_tensorfold_mtp_5",
      "model": "glm_4bit",
      "mode": "tensorfold_mtp",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 19.21318180975988,
      "metric": "Output token slope after token 32",
      "runtime_tps": 19.41243993571379,
      "ttft_seconds": 1228.4059184170328,
      "gpu_mhz": 397,
      "peak_load": 8.765625,
      "recall": 8,
      "qualified": false,
      "failed_checks": [
        "normal_decode_clock",
        "decode_clamp",
        "load"
      ],
      "quality_pass": true,
      "exact_pair": true,
      "kv_bits": null,
      "kv_group_size": null,
      "cache_format": "BF16 latent and indexer arrays; FP32 KDA state",
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "518ec9f645696a6b2f3f6331a553ad645c38c4c44f0ffb267f81cfd13e251066",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "The output bytes equal the manually reviewed full-context serial output.",
          "All 1024 tokens match the serial control.",
          "All eight answer records match the input.",
          "The same imprecise sentence-count description remains in the guide.",
          "This review does not validate every guide claim."
        ]
      },
      "setup_source": "glm53_tensorfold065_bench/results/run_20261005_145619/glm_4bit_260864_1234_tensorfold_mtp_5/result.json",
      "host_weight_preread_seconds": 1161.3357953340746,
      "host_weight_preread_bytes": 181709451790,
      "host_weight_preread_enabled": true,
      "startup_timeout_seconds": 3600,
      "setup_excluded_from_ttft": true,
      "eager_weight_load": false,
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "glm065",
      "source": "glm53_tensorfold065_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "glm53_tensorfold065_bench/results/run_20261005_145619/glm_4bit_260864_1234_tensorfold_mtp_7",
      "model": "glm_4bit",
      "mode": "tensorfold_cooled_prefill_mtp",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 44.54179006929351,
      "metric": "Output token slope after token 32",
      "runtime_tps": 45.46998252372363,
      "ttft_seconds": 1355.3322154581547,
      "gpu_mhz": 1380,
      "peak_load": 8.73046875,
      "recall": 8,
      "qualified": false,
      "failed_checks": [
        "load"
      ],
      "quality_pass": true,
      "exact_pair": true,
      "kv_bits": null,
      "kv_group_size": null,
      "cache_format": "BF16 latent and indexer arrays; FP32 KDA state",
      "cool_seconds": 251.0243173330091,
      "tokens_before_cool": 1,
      "prefill_pause_seconds": 0,
      "token_sha256": "518ec9f645696a6b2f3f6331a553ad645c38c4c44f0ffb267f81cfd13e251066",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "The answer bytes, prompt IDs, and all 1024 output tokens equal the reviewed serial pause control.",
          "All eight answer records match the full input.",
          "The output contains no visible repetition loop.",
          "The guide ends at the 1024-token output limit.",
          "The described Reykjavik context agrees with the nearby cistern sentences.",
          "The guide describes hundreds of sentences, although the prompt contains 13867 full stops.",
          "This review does not validate every guide claim."
        ]
      },
      "setup_source": "glm53_tensorfold065_bench/results/run_20261005_145619/glm_4bit_260864_1234_tensorfold_mtp_7/result.json",
      "host_weight_preread_seconds": 1917.6231765840203,
      "host_weight_preread_bytes": 181709451790,
      "host_weight_preread_enabled": true,
      "startup_timeout_seconds": 3600,
      "setup_excluded_from_ttft": true,
      "eager_weight_load": false,
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "glm065",
      "source": "glm53_tensorfold065_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "glm53_tensorfold065_bench/results/run_20261005_145619/glm_4bit_260864_1234_tensorfold_serial_4",
      "model": "glm_4bit",
      "mode": "tensorfold_serial",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 14.227255947224684,
      "metric": "Output token slope after token 32",
      "runtime_tps": 14.462180292525678,
      "ttft_seconds": 1328.4914443749003,
      "gpu_mhz": 338,
      "peak_load": 7.53759765625,
      "recall": 8,
      "qualified": false,
      "failed_checks": [
        "normal_decode_clock",
        "decode_clamp"
      ],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": null,
      "kv_group_size": null,
      "cache_format": "BF16 latent and indexer arrays; FP32 KDA state",
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "518ec9f645696a6b2f3f6331a553ad645c38c4c44f0ffb267f81cfd13e251066",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "All eight answer records match the full input.",
          "The output contains no visible repetition loop.",
          "The guide ends at the 1024-token output limit.",
          "The described Reykjavik context agrees with the nearby cistern sentences.",
          "The guide describes hundreds of sentences, although the prompt contains 13867 full stops.",
          "This review does not validate every guide claim."
        ]
      },
      "setup_source": "glm53_tensorfold065_bench/results/run_20261005_145619/glm_4bit_260864_1234_tensorfold_serial_4/result.json",
      "host_weight_preread_seconds": 14.766686625080183,
      "host_weight_preread_bytes": 181709451790,
      "host_weight_preread_enabled": true,
      "startup_timeout_seconds": 3600,
      "setup_excluded_from_ttft": true,
      "eager_weight_load": false,
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "glm065",
      "source": "glm53_tensorfold065_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "glm53_tensorfold065_bench/results/run_20261005_145619/glm_4bit_260864_1234_tensorfold_serial_6",
      "model": "glm_4bit",
      "mode": "tensorfold_cooled_prefill_serial",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 38.4858231880663,
      "metric": "Output token slope after token 32",
      "runtime_tps": 38.55320426786221,
      "ttft_seconds": 1580.5426103749778,
      "gpu_mhz": 1380,
      "peak_load": 4.6630859375,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": null,
      "kv_group_size": null,
      "cache_format": "BF16 latent and indexer arrays; FP32 KDA state",
      "cool_seconds": 255.88208433287218,
      "tokens_before_cool": 1,
      "prefill_pause_seconds": 0,
      "token_sha256": "518ec9f645696a6b2f3f6331a553ad645c38c4c44f0ffb267f81cfd13e251066",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "The answer bytes, prompt IDs, and output tokens equal the reviewed full-context serial control.",
          "All eight answer records match the full input.",
          "The output contains no visible repetition loop.",
          "The guide ends at the 1024-token output limit.",
          "The described Reykjavik context agrees with the nearby cistern sentences.",
          "The guide describes hundreds of sentences, although the prompt contains 13867 full stops.",
          "This review does not validate every guide claim."
        ]
      },
      "setup_source": "glm53_tensorfold065_bench/results/run_20261005_145619/glm_4bit_260864_1234_tensorfold_serial_6/result.json",
      "host_weight_preread_seconds": 439.8331142498646,
      "host_weight_preread_bytes": 181709451790,
      "host_weight_preread_enabled": true,
      "startup_timeout_seconds": 3600,
      "setup_excluded_from_ttft": true,
      "eager_weight_load": false,
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "glm065",
      "source": "glm53_tensorfold065_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "glm53_tensorfold065_bench/results/run_20261005_145619/glm_4bit_32768_1234_tensorfold_mtp_3",
      "model": "glm_4bit",
      "mode": "tensorfold_mtp",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 51.54127463455358,
      "metric": "Output token slope after token 32",
      "runtime_tps": 51.795246513853456,
      "ttft_seconds": 58.82565620820969,
      "gpu_mhz": 1380,
      "peak_load": 3.544921875,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": true,
      "kv_bits": null,
      "kv_group_size": null,
      "cache_format": "BF16 latent and indexer arrays; FP32 KDA state",
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "0b3d32910462349e7ec6be1a832dba8d27e1fc11420bdcfca9da8ef72f5cb859",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "The output bytes equal the manually reviewed serial output.",
          "All 1024 tokens match the serial control.",
          "All eight answer records match the input.",
          "The same incorrect position and context claims remain in the guide.",
          "The narrow retrieval pass does not validate the guide."
        ]
      },
      "setup_source": "glm53_tensorfold065_bench/results/run_20261005_145619/glm_4bit_32768_1234_tensorfold_mtp_3/result.json",
      "host_weight_preread_seconds": 15.053480708040297,
      "host_weight_preread_bytes": 181709451790,
      "host_weight_preread_enabled": true,
      "startup_timeout_seconds": 3600,
      "setup_excluded_from_ttft": true,
      "eager_weight_load": false,
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "glm065",
      "source": "glm53_tensorfold065_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "glm53_tensorfold065_bench/results/run_20261005_145619/glm_4bit_32768_1234_tensorfold_serial_2",
      "model": "glm_4bit",
      "mode": "tensorfold_serial",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 41.50942736121363,
      "metric": "Output token slope after token 32",
      "runtime_tps": 41.516976270107115,
      "ttft_seconds": 58.985859957989305,
      "gpu_mhz": 1380,
      "peak_load": 2.38623046875,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": null,
      "kv_group_size": null,
      "cache_format": "BF16 latent and indexer arrays; FP32 KDA state",
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "0b3d32910462349e7ec6be1a832dba8d27e1fc11420bdcfca9da8ef72f5cb859",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "All eight answer records match the input.",
          "The output contains no visible repetition loop.",
          "The guide ends at the 1024-token output limit.",
          "The guide places the Reykjavik record at one-fifth of the document. Its actual character position is about 2.9 percent.",
          "The guide gives two different descriptions of the Reykjavik context. Only the brittle-trellis description matches the input.",
          "The narrow retrieval pass does not validate the guide."
        ]
      },
      "setup_source": "glm53_tensorfold065_bench/results/run_20261005_145619/glm_4bit_32768_1234_tensorfold_serial_2/result.json",
      "host_weight_preread_seconds": 14.224005500087515,
      "host_weight_preread_bytes": 181709451790,
      "host_weight_preread_enabled": true,
      "startup_timeout_seconds": 3600,
      "setup_excluded_from_ttft": true,
      "eager_weight_load": false,
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "glm065",
      "source": "glm53_tensorfold065_bench/all-results.json",
      "status": "incomplete",
      "error": "RuntimeError('TensorFold exited with code 1')",
      "id": "glm53_tensorfold065_bench/results/run_20261005_145619/glm_oq4_2048_1234_tensorfold_serial_8",
      "model": "glm_oq4",
      "mode": "tensorfold_serial",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": null,
      "output_tokens": null,
      "cached_tokens": null,
      "decode_tps": null,
      "metric": "Output token slope after token 32",
      "runtime_tps": null,
      "ttft_seconds": null,
      "gpu_mhz": null,
      "peak_load": null,
      "recall": null,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": null,
      "exact_pair": null,
      "kv_bits": null,
      "kv_group_size": null,
      "cache_format": "BF16 latent and indexer arrays; FP32 KDA state",
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": null,
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": null,
      "setup_source": "glm53_tensorfold065_bench/results/run_20261005_145619/glm_oq4_2048_1234_tensorfold_serial_8/result.json",
      "host_weight_preread_seconds": 2213.5498648330104,
      "host_weight_preread_bytes": 185748648799,
      "host_weight_preread_enabled": true,
      "startup_timeout_seconds": 3600,
      "setup_excluded_from_ttft": true,
      "eager_weight_load": false,
      "startup_error": "RuntimeError: [METAL] Command buffer execution failed: Caused GPU Timeout Error (00000002:kIOGPUCommandBufferCallbackErrorTimeout).",
      "startup_failure_stage": "model_load",
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "glm065",
      "source": "glm53_tensorfold065_bench/all-results.json",
      "status": "incomplete",
      "error": "KeyboardInterrupt('signal 15')",
      "id": "glm53_tensorfold065_bench/results/run_20261005_200547/glm_oq4_2048_1234_tensorfold_mtp_1",
      "model": "glm_oq4",
      "mode": "tensorfold_mtp",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": null,
      "output_tokens": null,
      "cached_tokens": null,
      "decode_tps": null,
      "metric": "Output token slope after token 32",
      "runtime_tps": null,
      "ttft_seconds": null,
      "gpu_mhz": null,
      "peak_load": null,
      "recall": null,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": null,
      "exact_pair": null,
      "kv_bits": null,
      "kv_group_size": null,
      "cache_format": "BF16 latent and indexer arrays; FP32 KDA state",
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": null,
      "abort_reasons": [],
      "stop_reason": "The agent stopped the host read before the model server or measured request started.",
      "manual_review": null,
      "setup_source": "glm53_tensorfold065_bench/results/run_20261005_200547/glm_oq4_2048_1234_tensorfold_mtp_1/result.json",
      "host_weight_preread_seconds": null,
      "host_weight_preread_bytes": null,
      "host_weight_preread_enabled": true,
      "startup_timeout_seconds": 3600,
      "setup_excluded_from_ttft": true,
      "eager_weight_load": false,
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "glm065",
      "source": "glm53_tensorfold065_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "glm53_tensorfold065_bench/results/run_20261005_200547/glm_oq4_2048_1234_tensorfold_serial_0",
      "model": "glm_oq4",
      "mode": "tensorfold_serial",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 34.27510642444743,
      "metric": "Output token slope after token 32",
      "runtime_tps": 34.32646134315194,
      "ttft_seconds": 5.508592042140663,
      "gpu_mhz": 1380,
      "peak_load": 2.0478515625,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": null,
      "kv_group_size": null,
      "cache_format": "BF16 latent and indexer arrays; FP32 KDA state",
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "4f193f49081ab64437205b93e6c237a7b1b24ae87403ca7ac690b4146b342ac3",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "All eight answer records match the input, in the requested order.",
          "The output contains no visible repetition loop.",
          "The guide ends at the 1024-token output limit.",
          "The quoted distant-belfry text occurs at the start of the archive.",
          "The guide gives incorrect filler-sentence positions for Timbuktu, Kirkwall, Ushuaia, and Yakutsk.",
          "Their original filler-sentence positions are 14, 17, 20, and 24, rather than 15, 18, 22, and 26.",
          "This review does not validate every guide claim."
        ]
      },
      "setup_source": "glm53_tensorfold065_bench/results/run_20261005_200547/glm_oq4_2048_1234_tensorfold_serial_0/result.json",
      "host_weight_preread_seconds": 2082.4510087920353,
      "host_weight_preread_bytes": 185748648799,
      "host_weight_preread_enabled": true,
      "startup_timeout_seconds": 3600,
      "setup_excluded_from_ttft": true,
      "eager_weight_load": false,
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "glm065",
      "source": "glm53_tensorfold065_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "glm53_tensorfold065_bench/results/run_20261005_213646/glm_oq4_2048_1234_tensorfold_mtp_1",
      "model": "glm_oq4",
      "mode": "tensorfold_mtp",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 46.540066126448615,
      "metric": "Output token slope after token 32",
      "runtime_tps": 46.323845592911454,
      "ttft_seconds": 5.174186500022188,
      "gpu_mhz": 1380,
      "peak_load": 3.44775390625,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": true,
      "kv_bits": null,
      "kv_group_size": null,
      "cache_format": "BF16 latent and indexer arrays; FP32 KDA state",
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "4f193f49081ab64437205b93e6c237a7b1b24ae87403ca7ac690b4146b342ac3",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "The answer bytes, prompt IDs, and all 1024 output tokens equal the reviewed direct-load serial control.",
          "All eight answer records match the input, in the requested order.",
          "The output contains no visible repetition loop.",
          "The guide ends at the 1024-token output limit.",
          "The quoted distant-belfry text occurs at the start of the archive.",
          "The guide gives incorrect filler-sentence positions for Timbuktu, Kirkwall, Ushuaia, and Yakutsk.",
          "Their original filler-sentence positions are 14, 17, 20, and 24, rather than 15, 18, 22, and 26.",
          "This review does not validate every guide claim."
        ]
      },
      "setup_source": "glm53_tensorfold065_bench/results/run_20261005_213646/glm_oq4_2048_1234_tensorfold_mtp_1/result.json",
      "host_weight_preread_seconds": null,
      "host_weight_preread_bytes": null,
      "host_weight_preread_enabled": false,
      "startup_timeout_seconds": 3600,
      "setup_excluded_from_ttft": true,
      "eager_weight_load": false,
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "glm065",
      "source": "glm53_tensorfold065_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "glm53_tensorfold065_bench/results/run_20261005_213646/glm_oq4_2048_1234_tensorfold_serial_0",
      "model": "glm_oq4",
      "mode": "tensorfold_serial",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 34.50423971936668,
      "metric": "Output token slope after token 32",
      "runtime_tps": 34.50712886085122,
      "ttft_seconds": 5.210287083871663,
      "gpu_mhz": 1380,
      "peak_load": 3.06884765625,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": null,
      "kv_group_size": null,
      "cache_format": "BF16 latent and indexer arrays; FP32 KDA state",
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "4f193f49081ab64437205b93e6c237a7b1b24ae87403ca7ac690b4146b342ac3",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "The answer bytes, prompt IDs, and all 1024 output tokens equal the reviewed serial control with the separate host read.",
          "All eight answer records match the input, in the requested order.",
          "The output contains no visible repetition loop.",
          "The guide ends at the 1024-token output limit.",
          "The quoted distant-belfry text occurs at the start of the archive.",
          "The guide gives incorrect filler-sentence positions for Timbuktu, Kirkwall, Ushuaia, and Yakutsk.",
          "Their original filler-sentence positions are 14, 17, 20, and 24, rather than 15, 18, 22, and 26.",
          "This review does not validate every guide claim."
        ]
      },
      "setup_source": "glm53_tensorfold065_bench/results/run_20261005_213646/glm_oq4_2048_1234_tensorfold_serial_0/result.json",
      "host_weight_preread_seconds": null,
      "host_weight_preread_bytes": null,
      "host_weight_preread_enabled": false,
      "startup_timeout_seconds": 3600,
      "setup_excluded_from_ttft": true,
      "eager_weight_load": false,
      "reference_control": "glm53_tensorfold065_bench/results/run_20261005_200547/glm_oq4_2048_1234_tensorfold_serial_0",
      "reference_exact_tokens": true,
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "glm065",
      "source": "glm53_tensorfold065_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "glm53_tensorfold065_bench/results/run_20261005_213646/glm_oq4_260864_1234_tensorfold_mtp_5",
      "model": "glm_oq4",
      "mode": "tensorfold_mtp",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 14.593494724417877,
      "metric": "Output token slope after token 32",
      "runtime_tps": 14.480799053850223,
      "ttft_seconds": 1225.569438708946,
      "gpu_mhz": 343,
      "peak_load": 4.046875,
      "recall": 8,
      "qualified": false,
      "failed_checks": [
        "normal_decode_clock",
        "decode_clamp",
        "matched_serial_exactness"
      ],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": null,
      "kv_group_size": null,
      "cache_format": "BF16 latent and indexer arrays; FP32 KDA state",
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "9a439edb2d7a2064d380a3915d05a77b5e7b2ebb64aef406232d7ca95cc3d253",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "All eight answer records match the input, in the requested order.",
          "The output contains no visible repetition loop.",
          "The guide ends at the 1024-token output limit.",
          "Both quoted filler sentences occur in the input.",
          "The described Reykjavik context and its quoted continuation agree with the input.",
          "The guide starts a second record list as an extraction example.",
          "This review does not validate every guide claim.",
          "The baseline serial request did not complete, so the paired exactness check remains unavailable."
        ]
      },
      "exact_pair_unavailable_reason": "The serial control did not complete",
      "setup_source": "glm53_tensorfold065_bench/results/run_20261005_213646/glm_oq4_260864_1234_tensorfold_mtp_5/result.json",
      "host_weight_preread_seconds": null,
      "host_weight_preread_bytes": null,
      "host_weight_preread_enabled": false,
      "startup_timeout_seconds": 3600,
      "setup_excluded_from_ttft": true,
      "eager_weight_load": false,
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "glm065",
      "source": "glm53_tensorfold065_bench/all-results.json",
      "status": "incomplete",
      "error": "KeyboardInterrupt('signal 15')",
      "id": "glm53_tensorfold065_bench/results/run_20261005_213646/glm_oq4_260864_1234_tensorfold_serial_4",
      "model": "glm_oq4",
      "mode": "tensorfold_serial",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": null,
      "output_tokens": null,
      "cached_tokens": null,
      "decode_tps": null,
      "metric": "Output token slope after token 32",
      "runtime_tps": null,
      "ttft_seconds": null,
      "gpu_mhz": null,
      "peak_load": 26.62109375,
      "recall": null,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": null,
      "exact_pair": null,
      "kv_bits": null,
      "kv_group_size": null,
      "cache_format": "BF16 latent and indexer arrays; FP32 KDA state",
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": null,
      "abort_reasons": [
        "System load exceeded16 twice"
      ],
      "stop_reason": null,
      "manual_review": null,
      "setup_source": "glm53_tensorfold065_bench/results/run_20261005_213646/glm_oq4_260864_1234_tensorfold_serial_4/result.json",
      "host_weight_preread_seconds": null,
      "host_weight_preread_bytes": null,
      "host_weight_preread_enabled": false,
      "startup_timeout_seconds": 3600,
      "setup_excluded_from_ttft": true,
      "eager_weight_load": false,
      "last_prefill_tokens": 131072,
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "glm065",
      "source": "glm53_tensorfold065_bench/all-results.json",
      "status": "incomplete",
      "error": "RuntimeError('TensorFold exited with code 1')",
      "id": "glm53_tensorfold065_bench/results/run_20261005_213646/glm_oq4_260864_1234_tensorfold_serial_6",
      "model": "glm_oq4",
      "mode": "tensorfold_cooled_prefill_serial",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": null,
      "output_tokens": null,
      "cached_tokens": null,
      "decode_tps": null,
      "metric": "Output token slope after token 32",
      "runtime_tps": null,
      "ttft_seconds": null,
      "gpu_mhz": null,
      "peak_load": null,
      "recall": null,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": null,
      "exact_pair": null,
      "kv_bits": null,
      "kv_group_size": null,
      "cache_format": "BF16 latent and indexer arrays; FP32 KDA state",
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": null,
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": null,
      "setup_source": "glm53_tensorfold065_bench/results/run_20261005_213646/glm_oq4_260864_1234_tensorfold_serial_6/result.json",
      "host_weight_preread_seconds": null,
      "host_weight_preread_bytes": null,
      "host_weight_preread_enabled": false,
      "startup_timeout_seconds": 3600,
      "setup_excluded_from_ttft": true,
      "eager_weight_load": false,
      "startup_error": "RuntimeError: [METAL] Command buffer execution failed: Caused GPU Timeout Error (00000002:kIOGPUCommandBufferCallbackErrorTimeout).",
      "startup_failure_stage": "model_load",
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "glm065",
      "source": "glm53_tensorfold065_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "glm53_tensorfold065_bench/results/run_20261005_213646/glm_oq4_32768_1234_tensorfold_mtp_3",
      "model": "glm_oq4",
      "mode": "tensorfold_mtp",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 46.66492572403743,
      "metric": "Output token slope after token 32",
      "runtime_tps": 46.373977312413544,
      "ttft_seconds": 60.88952837511897,
      "gpu_mhz": 1380,
      "peak_load": 2.47412109375,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": true,
      "kv_bits": null,
      "kv_group_size": null,
      "cache_format": "BF16 latent and indexer arrays; FP32 KDA state",
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "268da586fa408204a35caa3b0973759e30ecddf6ff52e3b70b9061f7109da670",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "The answer bytes, prompt IDs, and all 1024 output tokens equal the reviewed serial control.",
          "All eight answer records match the input, in the requested order.",
          "The output contains no visible repetition loop.",
          "The guide ends at the 1024-token output limit.",
          "The quoted Reykjavik sentence appears verbatim in the input.",
          "The guide describes hundreds of sentences, although the prompt contains 1680 full stops.",
          "This review does not validate every guide claim."
        ]
      },
      "setup_source": "glm53_tensorfold065_bench/results/run_20261005_213646/glm_oq4_32768_1234_tensorfold_mtp_3/result.json",
      "host_weight_preread_seconds": null,
      "host_weight_preread_bytes": null,
      "host_weight_preread_enabled": false,
      "startup_timeout_seconds": 3600,
      "setup_excluded_from_ttft": true,
      "eager_weight_load": false,
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "glm065",
      "source": "glm53_tensorfold065_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "glm53_tensorfold065_bench/results/run_20261005_213646/glm_oq4_32768_1234_tensorfold_serial_2",
      "model": "glm_oq4",
      "mode": "tensorfold_serial",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 33.76843345363697,
      "metric": "Output token slope after token 32",
      "runtime_tps": 33.76791102659878,
      "ttft_seconds": 61.04101283312775,
      "gpu_mhz": 1380,
      "peak_load": 3.83203125,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": null,
      "kv_group_size": null,
      "cache_format": "BF16 latent and indexer arrays; FP32 KDA state",
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "268da586fa408204a35caa3b0973759e30ecddf6ff52e3b70b9061f7109da670",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "All eight answer records match the input, in the requested order.",
          "The output contains no visible repetition loop.",
          "The guide ends at the 1024-token output limit.",
          "The quoted Reykjavik sentence appears verbatim in the input.",
          "The guide describes hundreds of sentences, although the prompt contains 1680 full stops.",
          "This review does not validate every guide claim."
        ]
      },
      "setup_source": "glm53_tensorfold065_bench/results/run_20261005_213646/glm_oq4_32768_1234_tensorfold_serial_2/result.json",
      "host_weight_preread_seconds": null,
      "host_weight_preread_bytes": null,
      "host_weight_preread_enabled": false,
      "startup_timeout_seconds": 3600,
      "setup_excluded_from_ttft": true,
      "eager_weight_load": false,
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "glm065",
      "source": "glm53_tensorfold065_bench/all-results.json",
      "status": "incomplete",
      "error": "RuntimeError('TensorFold exited with code 1')",
      "id": "glm53_tensorfold065_bench/results/run_20261005_230806/glm_oq4_260864_1234_tensorfold_serial_0",
      "model": "glm_oq4",
      "mode": "tensorfold_serial",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": null,
      "output_tokens": null,
      "cached_tokens": null,
      "decode_tps": null,
      "metric": "Output token slope after token 32",
      "runtime_tps": null,
      "ttft_seconds": null,
      "gpu_mhz": null,
      "peak_load": null,
      "recall": null,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": null,
      "exact_pair": null,
      "kv_bits": null,
      "kv_group_size": null,
      "cache_format": "BF16 latent and indexer arrays; FP32 KDA state",
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": null,
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": null,
      "setup_source": "glm53_tensorfold065_bench/results/run_20261005_230806/glm_oq4_260864_1234_tensorfold_serial_0/result.json",
      "host_weight_preread_seconds": null,
      "host_weight_preread_bytes": null,
      "host_weight_preread_enabled": false,
      "startup_timeout_seconds": 3600,
      "setup_excluded_from_ttft": true,
      "eager_weight_load": false,
      "startup_error": "RuntimeError: [METAL] Command buffer execution failed: Caused GPU Timeout Error (00000002:kIOGPUCommandBufferCallbackErrorTimeout).",
      "startup_failure_stage": "model_load",
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "glm065",
      "source": "glm53_tensorfold065_bench/all-results.json",
      "status": "incomplete",
      "error": "RuntimeError('TensorFold exited with code 1')",
      "id": "glm53_tensorfold065_bench/results/run_20261005_232901/glm_oq2_2048_1234_tensorfold_serial_0",
      "model": "glm_oq2",
      "mode": "tensorfold_serial",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": null,
      "output_tokens": null,
      "cached_tokens": null,
      "decode_tps": null,
      "metric": "Output token slope after token 32",
      "runtime_tps": null,
      "ttft_seconds": null,
      "gpu_mhz": null,
      "peak_load": null,
      "recall": null,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": null,
      "exact_pair": null,
      "kv_bits": null,
      "kv_group_size": null,
      "cache_format": "BF16 latent and indexer arrays; FP32 KDA state",
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": null,
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": null,
      "setup_source": "glm53_tensorfold065_bench/results/run_20261005_232901/glm_oq2_2048_1234_tensorfold_serial_0/result.json",
      "host_weight_preread_seconds": null,
      "host_weight_preread_bytes": null,
      "host_weight_preread_enabled": false,
      "startup_timeout_seconds": 3600,
      "setup_excluded_from_ttft": true,
      "eager_weight_load": false,
      "startup_error": "RuntimeError: [METAL] Command buffer execution failed: Caused GPU Timeout Error (00000002:kIOGPUCommandBufferCallbackErrorTimeout).",
      "startup_failure_stage": "model_load",
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "glm065",
      "source": "glm53_tensorfold065_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "glm53_tensorfold065_bench/results/run_20261005_234048/glm_oq4_2048_1234_tensorfold_serial_0",
      "model": "glm_oq4",
      "mode": "tensorfold_eager_load_serial",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 34.53889242062779,
      "metric": "Output token slope after token 32",
      "runtime_tps": 34.53015586700828,
      "ttft_seconds": 5.681261624908075,
      "gpu_mhz": 1380,
      "peak_load": 5.728515625,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": null,
      "kv_group_size": null,
      "cache_format": "BF16 latent and indexer arrays; FP32 KDA state",
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "4f193f49081ab64437205b93e6c237a7b1b24ae87403ca7ac690b4146b342ac3",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "The answer bytes, prompt IDs, and all 1024 output tokens equal the reviewed serial control with the separate host read.",
          "All eight answer records match the input, in the requested order.",
          "The output contains no visible repetition loop.",
          "The guide ends at the 1024-token output limit.",
          "The quoted distant-belfry text occurs at the start of the archive.",
          "The guide gives incorrect filler-sentence positions for Timbuktu, Kirkwall, Ushuaia, and Yakutsk.",
          "Their original filler-sentence positions are 14, 17, 20, and 24, rather than 15, 18, 22, and 26.",
          "This review does not validate every guide claim."
        ]
      },
      "setup_source": "glm53_tensorfold065_bench/results/run_20261005_234048/glm_oq4_2048_1234_tensorfold_serial_0/result.json",
      "host_weight_preread_seconds": null,
      "host_weight_preread_bytes": null,
      "host_weight_preread_enabled": false,
      "startup_timeout_seconds": 3600,
      "setup_excluded_from_ttft": true,
      "eager_weight_load": true,
      "loader_setup_seconds": 3261.6066300831735,
      "loader_variant": {
        "variant": "eager_array_load",
        "completed": true,
        "hooks_restored": true,
        "seconds": 3261.6066300831735,
        "calls": 2778,
        "source_sha256": "5bcb0ec46a185d500aa65635dc034882059d37164b2b858b135c4f8e4e7186d6"
      },
      "reference_control": "glm53_tensorfold065_bench/results/run_20261005_213646/glm_oq4_2048_1234_tensorfold_serial_0",
      "reference_exact_tokens": true,
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "glm065",
      "source": "glm53_tensorfold065_bench/all-results.json",
      "status": "incomplete",
      "error": "KeyboardInterrupt('signal 15')",
      "id": "glm53_tensorfold065_bench/results/run_20261005_234048/glm_oq4_260864_1234_tensorfold_serial_2",
      "model": "glm_oq4",
      "mode": "tensorfold_eager_load_serial",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": null,
      "output_tokens": null,
      "cached_tokens": null,
      "decode_tps": null,
      "metric": "Output token slope after token 32",
      "runtime_tps": null,
      "ttft_seconds": null,
      "gpu_mhz": null,
      "peak_load": 6.42431640625,
      "recall": null,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": null,
      "exact_pair": null,
      "kv_bits": null,
      "kv_group_size": null,
      "cache_format": "BF16 latent and indexer arrays; FP32 KDA state",
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": null,
      "abort_reasons": [],
      "stop_reason": "The driver received SIGHUP after the assistant service restart. Its cleanup sent SIGTERM to the active campaign and worker.",
      "manual_review": null,
      "setup_source": "glm53_tensorfold065_bench/results/run_20261005_234048/glm_oq4_260864_1234_tensorfold_serial_2/result.json",
      "host_weight_preread_seconds": null,
      "host_weight_preread_bytes": null,
      "host_weight_preread_enabled": false,
      "startup_timeout_seconds": 3600,
      "setup_excluded_from_ttft": true,
      "eager_weight_load": true,
      "loader_setup_seconds": 3266.540320875123,
      "loader_variant": {
        "variant": "eager_array_load",
        "completed": true,
        "hooks_restored": true,
        "seconds": 3266.540320875123,
        "calls": 2778,
        "source_sha256": "5bcb0ec46a185d500aa65635dc034882059d37164b2b858b135c4f8e4e7186d6"
      },
      "last_prefill_tokens": 212992,
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "glm065",
      "source": "glm53_tensorfold065_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "glm53_tensorfold065_bench/results/run_20261005_234048/glm_oq4_32768_1234_tensorfold_serial_1",
      "model": "glm_oq4",
      "mode": "tensorfold_eager_load_serial",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 33.69778656454137,
      "metric": "Output token slope after token 32",
      "runtime_tps": 33.696758287547134,
      "ttft_seconds": 61.1505509999115,
      "gpu_mhz": 1380,
      "peak_load": 2.89794921875,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": null,
      "kv_group_size": null,
      "cache_format": "BF16 latent and indexer arrays; FP32 KDA state",
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "268da586fa408204a35caa3b0973759e30ecddf6ff52e3b70b9061f7109da670",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "All eight answer records match the input, in the requested order.",
          "The output contains no visible repetition loop.",
          "The guide ends at the 1024-token output limit.",
          "The quoted Reykjavik sentence appears verbatim in the input.",
          "The guide describes hundreds of sentences, although the prompt contains 1680 full stops.",
          "This review does not validate every guide claim."
        ]
      },
      "setup_source": "glm53_tensorfold065_bench/results/run_20261005_234048/glm_oq4_32768_1234_tensorfold_serial_1/result.json",
      "host_weight_preread_seconds": null,
      "host_weight_preread_bytes": null,
      "host_weight_preread_enabled": false,
      "startup_timeout_seconds": 3600,
      "setup_excluded_from_ttft": true,
      "eager_weight_load": true,
      "loader_setup_seconds": 2883.1688115000725,
      "loader_variant": {
        "variant": "eager_array_load",
        "completed": true,
        "hooks_restored": true,
        "seconds": 2883.1688115000725,
        "calls": 2778,
        "source_sha256": "5bcb0ec46a185d500aa65635dc034882059d37164b2b858b135c4f8e4e7186d6"
      },
      "reference_control": "glm53_tensorfold065_bench/results/run_20261005_213646/glm_oq4_32768_1234_tensorfold_serial_2",
      "reference_exact_tokens": true,
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "glm065",
      "source": "glm53_tensorfold065_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "glm53_tensorfold065_bench/results/run_20261006_025508/glm_oq2_2048_1234_tensorfold_mtp_3",
      "model": "glm_oq2",
      "mode": "tensorfold_eager_load_mtp",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 39.85535468767252,
      "metric": "Output token slope after token 32",
      "runtime_tps": 40.427911928223615,
      "ttft_seconds": 4.722372000105679,
      "gpu_mhz": 1380,
      "peak_load": 2.23486328125,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": true,
      "kv_bits": null,
      "kv_group_size": null,
      "cache_format": "BF16 latent and indexer arrays; FP32 KDA state",
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "5c181fcf87bd47a24285b1908b1055f879579d659f65d15ae2f7385e3e70f3e9",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "All eight answer records match the input, in the requested order.",
          "The output contains no visible repetition loop.",
          "The quoted opening fragment occurs in the input.",
          "The Reykjavik example correctly identifies the first record and its immediate context.",
          "The guide ends at the 1024-token output limit.",
          "This review does not validate every guide claim."
        ]
      },
      "setup_source": "glm53_tensorfold065_bench/results/run_20261006_025508/glm_oq2_2048_1234_tensorfold_mtp_3/result.json",
      "host_weight_preread_seconds": null,
      "host_weight_preread_bytes": null,
      "host_weight_preread_enabled": false,
      "startup_timeout_seconds": 3600,
      "setup_excluded_from_ttft": true,
      "eager_weight_load": true,
      "loader_setup_seconds": 1105.6067729999777,
      "loader_variant": {
        "variant": "eager_array_load",
        "completed": true,
        "hooks_restored": true,
        "seconds": 1105.6067729999777,
        "calls": 2778,
        "source_sha256": "5bcb0ec46a185d500aa65635dc034882059d37164b2b858b135c4f8e4e7186d6"
      },
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "glm065",
      "source": "glm53_tensorfold065_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "glm53_tensorfold065_bench/results/run_20261006_025508/glm_oq2_2048_1234_tensorfold_serial_2",
      "model": "glm_oq2",
      "mode": "tensorfold_eager_load_serial",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 31.06437802444768,
      "metric": "Output token slope after token 32",
      "runtime_tps": 31.066011587127488,
      "ttft_seconds": 5.3045776670333,
      "gpu_mhz": 1380,
      "peak_load": 2.64892578125,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": null,
      "kv_group_size": null,
      "cache_format": "BF16 latent and indexer arrays; FP32 KDA state",
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "5c181fcf87bd47a24285b1908b1055f879579d659f65d15ae2f7385e3e70f3e9",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "All eight answer records match the input, in the requested order.",
          "The output contains no visible repetition loop.",
          "The quoted opening fragment occurs in the input.",
          "The Reykjavik example correctly identifies the first record and its immediate context.",
          "The guide ends at the 1024-token output limit.",
          "This review does not validate every guide claim."
        ]
      },
      "setup_source": "glm53_tensorfold065_bench/results/run_20261006_025508/glm_oq2_2048_1234_tensorfold_serial_2/result.json",
      "host_weight_preread_seconds": null,
      "host_weight_preread_bytes": null,
      "host_weight_preread_enabled": false,
      "startup_timeout_seconds": 3600,
      "setup_excluded_from_ttft": true,
      "eager_weight_load": true,
      "loader_setup_seconds": 2034.8739753749687,
      "loader_variant": {
        "variant": "eager_array_load",
        "completed": true,
        "hooks_restored": true,
        "seconds": 2034.8739753749687,
        "calls": 2778,
        "source_sha256": "5bcb0ec46a185d500aa65635dc034882059d37164b2b858b135c4f8e4e7186d6"
      },
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "glm065",
      "source": "glm53_tensorfold065_bench/all-results.json",
      "status": "incomplete",
      "error": "KeyboardInterrupt('signal 15')",
      "id": "glm53_tensorfold065_bench/results/run_20261006_025508/glm_oq2_260864_1234_tensorfold_mtp_7",
      "model": "glm_oq2",
      "mode": "tensorfold_eager_load_mtp",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": null,
      "output_tokens": null,
      "cached_tokens": null,
      "decode_tps": null,
      "metric": "Output token slope after token 32",
      "runtime_tps": null,
      "ttft_seconds": null,
      "gpu_mhz": null,
      "peak_load": 20.06591796875,
      "recall": null,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": null,
      "exact_pair": null,
      "kv_bits": null,
      "kv_group_size": null,
      "cache_format": "BF16 latent and indexer arrays; FP32 KDA state",
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": null,
      "abort_reasons": [
        "System load exceeded16 twice"
      ],
      "stop_reason": null,
      "manual_review": null,
      "setup_source": "glm53_tensorfold065_bench/results/run_20261006_025508/glm_oq2_260864_1234_tensorfold_mtp_7/result.json",
      "host_weight_preread_seconds": null,
      "host_weight_preread_bytes": null,
      "host_weight_preread_enabled": false,
      "startup_timeout_seconds": 3600,
      "setup_excluded_from_ttft": true,
      "eager_weight_load": true,
      "loader_setup_seconds": 71.92580462503247,
      "loader_variant": {
        "variant": "eager_array_load",
        "completed": true,
        "hooks_restored": true,
        "seconds": 71.92580462503247,
        "calls": 2778,
        "source_sha256": "5bcb0ec46a185d500aa65635dc034882059d37164b2b858b135c4f8e4e7186d6"
      },
      "last_prefill_tokens": 114688,
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "glm065",
      "source": "glm53_tensorfold065_bench/all-results.json",
      "status": "incomplete",
      "error": "KeyboardInterrupt('signal 15')",
      "id": "glm53_tensorfold065_bench/results/run_20261006_025508/glm_oq2_260864_1234_tensorfold_serial_6",
      "model": "glm_oq2",
      "mode": "tensorfold_eager_load_serial",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": null,
      "output_tokens": null,
      "cached_tokens": null,
      "decode_tps": null,
      "metric": "Output token slope after token 32",
      "runtime_tps": null,
      "ttft_seconds": null,
      "gpu_mhz": null,
      "peak_load": 19.9033203125,
      "recall": null,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": null,
      "exact_pair": null,
      "kv_bits": null,
      "kv_group_size": null,
      "cache_format": "BF16 latent and indexer arrays; FP32 KDA state",
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": null,
      "abort_reasons": [
        "System load exceeded16 twice"
      ],
      "stop_reason": null,
      "manual_review": null,
      "setup_source": "glm53_tensorfold065_bench/results/run_20261006_025508/glm_oq2_260864_1234_tensorfold_serial_6/result.json",
      "host_weight_preread_seconds": null,
      "host_weight_preread_bytes": null,
      "host_weight_preread_enabled": false,
      "startup_timeout_seconds": 3600,
      "setup_excluded_from_ttft": true,
      "eager_weight_load": true,
      "loader_setup_seconds": 46.87106633395888,
      "loader_variant": {
        "variant": "eager_array_load",
        "completed": true,
        "hooks_restored": true,
        "seconds": 46.87106633395888,
        "calls": 2778,
        "source_sha256": "5bcb0ec46a185d500aa65635dc034882059d37164b2b858b135c4f8e4e7186d6"
      },
      "last_prefill_tokens": 114688,
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "glm065",
      "source": "glm53_tensorfold065_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "glm53_tensorfold065_bench/results/run_20261006_025508/glm_oq2_32768_1234_tensorfold_mtp_5",
      "model": "glm_oq2",
      "mode": "tensorfold_eager_load_mtp",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 46.210978222287885,
      "metric": "Output token slope after token 32",
      "runtime_tps": 47.406453041658466,
      "ttft_seconds": 64.91792866704054,
      "gpu_mhz": 1380,
      "peak_load": 3.32080078125,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": true,
      "kv_bits": null,
      "kv_group_size": null,
      "cache_format": "BF16 latent and indexer arrays; FP32 KDA state",
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "9fef67a04af669819bc41b4328eba92491dd118e167fbd358926a97e12f57bbc",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "All eight answer records match the input, in the requested order.",
          "The output contains no visible repetition loop.",
          "The five complete quoted record examples match the input.",
          "The partial Kirkwall example also matches the input before the output ends.",
          "The Timbuktu record occurs after the complete phrase faded near; it does not split that phrase itself.",
          "The stated common insertion position does not describe every record.",
          "The guide says hundreds of sentences; the archive contains 1,669 full stops.",
          "The guide ends at the 1024-token output limit.",
          "This review does not validate every guide claim."
        ]
      },
      "setup_source": "glm53_tensorfold065_bench/results/run_20261006_025508/glm_oq2_32768_1234_tensorfold_mtp_5/result.json",
      "host_weight_preread_seconds": null,
      "host_weight_preread_bytes": null,
      "host_weight_preread_enabled": false,
      "startup_timeout_seconds": 3600,
      "setup_excluded_from_ttft": true,
      "eager_weight_load": true,
      "loader_setup_seconds": 236.83042912511155,
      "loader_variant": {
        "variant": "eager_array_load",
        "completed": true,
        "hooks_restored": true,
        "seconds": 236.83042912511155,
        "calls": 2778,
        "source_sha256": "5bcb0ec46a185d500aa65635dc034882059d37164b2b858b135c4f8e4e7186d6"
      },
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "glm065",
      "source": "glm53_tensorfold065_bench/all-results.json",
      "status": "completed",
      "error": null,
      "id": "glm53_tensorfold065_bench/results/run_20261006_025508/glm_oq2_32768_1234_tensorfold_serial_4",
      "model": "glm_oq2",
      "mode": "tensorfold_eager_load_serial",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 30.53532399740006,
      "metric": "Output token slope after token 32",
      "runtime_tps": 30.548959615511503,
      "ttft_seconds": 64.92808812484145,
      "gpu_mhz": 1380,
      "peak_load": 3.80810546875,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "quality_pass": true,
      "exact_pair": null,
      "kv_bits": null,
      "kv_group_size": null,
      "cache_format": "BF16 latent and indexer arrays; FP32 KDA state",
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": "9fef67a04af669819bc41b4328eba92491dd118e167fbd358926a97e12f57bbc",
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": {
        "retrieval_records_correct": true,
        "visible_repetition_loop": false,
        "narrow_output_pass": true,
        "general_quality_validated": false,
        "notes": [
          "All eight answer records match the input, in the requested order.",
          "The output contains no visible repetition loop.",
          "The five complete quoted record examples match the input.",
          "The partial Kirkwall example also matches the input before the output ends.",
          "The Timbuktu record occurs after the complete phrase faded near; it does not split that phrase itself.",
          "The stated common insertion position does not describe every record.",
          "The guide says hundreds of sentences; the archive contains 1,669 full stops.",
          "The guide ends at the 1024-token output limit.",
          "This review does not validate every guide claim."
        ]
      },
      "setup_source": "glm53_tensorfold065_bench/results/run_20261006_025508/glm_oq2_32768_1234_tensorfold_serial_4/result.json",
      "host_weight_preread_seconds": null,
      "host_weight_preread_bytes": null,
      "host_weight_preread_enabled": false,
      "startup_timeout_seconds": 3600,
      "setup_excluded_from_ttft": true,
      "eager_weight_load": true,
      "loader_setup_seconds": 161.95395129104145,
      "loader_variant": {
        "variant": "eager_array_load",
        "completed": true,
        "hooks_restored": true,
        "seconds": 161.95395129104145,
        "calls": 2778,
        "source_sha256": "5bcb0ec46a185d500aa65635dc034882059d37164b2b858b135c4f8e4e7186d6"
      },
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "glm065",
      "source": "glm53_tensorfold065_bench/all-results.json",
      "status": "incomplete",
      "error": "KeyboardInterrupt('signal 15')",
      "id": "glm53_tensorfold065_bench/results/run_20261006_025508/glm_oq4_260864_1234_tensorfold_mtp_1",
      "model": "glm_oq4",
      "mode": "tensorfold_eager_load_mtp",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": null,
      "output_tokens": null,
      "cached_tokens": null,
      "decode_tps": null,
      "metric": "Output token slope after token 32",
      "runtime_tps": null,
      "ttft_seconds": null,
      "gpu_mhz": null,
      "peak_load": 25.90478515625,
      "recall": null,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": null,
      "exact_pair": null,
      "kv_bits": null,
      "kv_group_size": null,
      "cache_format": "BF16 latent and indexer arrays; FP32 KDA state",
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": null,
      "abort_reasons": [
        "System load exceeded16 twice"
      ],
      "stop_reason": null,
      "manual_review": null,
      "setup_source": "glm53_tensorfold065_bench/results/run_20261006_025508/glm_oq4_260864_1234_tensorfold_mtp_1/result.json",
      "host_weight_preread_seconds": null,
      "host_weight_preread_bytes": null,
      "host_weight_preread_enabled": false,
      "startup_timeout_seconds": 3600,
      "setup_excluded_from_ttft": true,
      "eager_weight_load": true,
      "loader_setup_seconds": 2308.6601259999443,
      "loader_variant": {
        "variant": "eager_array_load",
        "completed": true,
        "hooks_restored": true,
        "seconds": 2308.6601259999443,
        "calls": 2778,
        "source_sha256": "5bcb0ec46a185d500aa65635dc034882059d37164b2b858b135c4f8e4e7186d6"
      },
      "last_prefill_tokens": 163840,
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "glm065",
      "source": "glm53_tensorfold065_bench/all-results.json",
      "status": "incomplete",
      "error": "KeyboardInterrupt('signal 15')",
      "id": "glm53_tensorfold065_bench/results/run_20261006_025508/glm_oq4_260864_1234_tensorfold_serial_0",
      "model": "glm_oq4",
      "mode": "tensorfold_eager_load_serial",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": null,
      "output_tokens": null,
      "cached_tokens": null,
      "decode_tps": null,
      "metric": "Output token slope after token 32",
      "runtime_tps": null,
      "ttft_seconds": null,
      "gpu_mhz": null,
      "peak_load": 21.82080078125,
      "recall": null,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": null,
      "exact_pair": null,
      "kv_bits": null,
      "kv_group_size": null,
      "cache_format": "BF16 latent and indexer arrays; FP32 KDA state",
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": null,
      "abort_reasons": [
        "System load exceeded16 twice"
      ],
      "stop_reason": null,
      "manual_review": null,
      "setup_source": "glm53_tensorfold065_bench/results/run_20261006_025508/glm_oq4_260864_1234_tensorfold_serial_0/result.json",
      "host_weight_preread_seconds": null,
      "host_weight_preread_bytes": null,
      "host_weight_preread_enabled": false,
      "startup_timeout_seconds": 3600,
      "setup_excluded_from_ttft": true,
      "eager_weight_load": true,
      "loader_setup_seconds": 2445.672217708081,
      "loader_variant": {
        "variant": "eager_array_load",
        "completed": true,
        "hooks_restored": true,
        "seconds": 2445.672217708081,
        "calls": 2778,
        "source_sha256": "5bcb0ec46a185d500aa65635dc034882059d37164b2b858b135c4f8e4e7186d6"
      },
      "last_prefill_tokens": 163840,
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "glm065",
      "source": "glm53_tensorfold065_bench/all-results.json",
      "status": "incomplete",
      "error": "TimeoutError('TensorFold startup exceeded 3600 seconds')",
      "id": "glm53_tensorfold065_bench/results/run_20261006_063141/glm_oq4_260864_1234_tensorfold_serial_0",
      "model": "glm_oq4",
      "mode": "tensorfold_eager_load_cooled_prefill_serial",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": null,
      "output_tokens": null,
      "cached_tokens": null,
      "decode_tps": null,
      "metric": "Output token slope after token 32",
      "runtime_tps": null,
      "ttft_seconds": null,
      "gpu_mhz": null,
      "peak_load": null,
      "recall": null,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": null,
      "exact_pair": null,
      "kv_bits": null,
      "kv_group_size": null,
      "cache_format": "BF16 latent and indexer arrays; FP32 KDA state",
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": null,
      "abort_reasons": [],
      "stop_reason": null,
      "manual_review": null,
      "setup_source": "glm53_tensorfold065_bench/results/run_20261006_063141/glm_oq4_260864_1234_tensorfold_serial_0/result.json",
      "host_weight_preread_seconds": null,
      "host_weight_preread_bytes": null,
      "host_weight_preread_enabled": false,
      "startup_timeout_seconds": 3600,
      "setup_excluded_from_ttft": true,
      "eager_weight_load": true,
      "startup_error": "TimeoutError('TensorFold startup exceeded 3600 seconds')",
      "startup_failure_stage": "model_load",
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "glm065",
      "source": "glm53_tensorfold065_bench/all-results.json",
      "status": "incomplete",
      "error": "KeyboardInterrupt('signal 15')",
      "id": "glm53_tensorfold065_bench/results/run_20261006_073714/glm_oq4_260864_1234_tensorfold_mtp_0",
      "model": "glm_oq4",
      "mode": "tensorfold_eager_load_cooled_prefill_mtp",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": null,
      "output_tokens": null,
      "cached_tokens": null,
      "decode_tps": null,
      "metric": "Output token slope after token 32",
      "runtime_tps": null,
      "ttft_seconds": null,
      "gpu_mhz": null,
      "peak_load": 20.7294921875,
      "recall": null,
      "qualified": false,
      "failed_checks": [],
      "quality_pass": null,
      "exact_pair": null,
      "kv_bits": null,
      "kv_group_size": null,
      "cache_format": "BF16 latent and indexer arrays; FP32 KDA state",
      "cool_seconds": 0,
      "tokens_before_cool": 0,
      "prefill_pause_seconds": 0,
      "token_sha256": null,
      "abort_reasons": [
        "System load exceeded16 twice"
      ],
      "stop_reason": null,
      "manual_review": null,
      "setup_source": "glm53_tensorfold065_bench/results/run_20261006_073714/glm_oq4_260864_1234_tensorfold_mtp_0/result.json",
      "host_weight_preread_seconds": null,
      "host_weight_preread_bytes": null,
      "host_weight_preread_enabled": false,
      "startup_timeout_seconds": 3600,
      "setup_excluded_from_ttft": true,
      "eager_weight_load": true,
      "loader_setup_seconds": 2264.202658791095,
      "loader_variant": {
        "variant": "eager_array_load",
        "completed": true,
        "hooks_restored": true,
        "seconds": 2264.202658791095,
        "calls": 2778,
        "source_sha256": "5bcb0ec46a185d500aa65635dc034882059d37164b2b858b135c4f8e4e7186d6"
      },
      "background_intervention": {
        "at": "2026-10-06T00:27:48.219087+00:00",
        "action": "terminate",
        "authorized": true,
        "process_count": 2,
        "measured_request_active": true,
        "last_completed_prefill_tokens": 163840,
        "processes_ended": true,
        "note": "The user authorized both terminations during this request. This change prevents a comparison under identical background conditions."
      },
      "last_prefill_tokens": 163840,
      "runtime_id": "tensorfold065"
    },
    {
      "campaign": "glm",
      "source": "glm53_bench/summary.json",
      "status": "completed",
      "error": null,
      "id": "glm53_bench/oq2/results/run_20260930_135359/256k_retrieval_1234_draft",
      "model": "glm53_oq2",
      "mode": "MTP + context copies",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 17.2623914537536,
      "metric": "Client token count and first/last text arrival",
      "runtime_tps": 17.27129451774994,
      "ttft_seconds": 2178.455887334072,
      "gpu_mhz": 425,
      "recall": 8,
      "qualified": false,
      "failed_checks": [
        "median decode clock >=1300 MHz",
        "overall clamp <=5%",
        "decode clamp <=5%",
        "load <=8"
      ],
      "exact_pair": true,
      "token_sha256": "5a0bf55103ed",
      "cool_seconds": 0,
      "runtime_id": "tensorfold0363_glm"
    },
    {
      "campaign": "glm",
      "source": "glm53_bench/summary.json",
      "status": "completed",
      "error": null,
      "id": "glm53_bench/oq2/results/run_20260930_135359/256k_retrieval_1234_serial",
      "model": "glm53_oq2",
      "mode": "Serial",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 12.185071819856912,
      "metric": "Client token count and first/last text arrival",
      "runtime_tps": 12.190936339734883,
      "ttft_seconds": 2268.308848665911,
      "gpu_mhz": 339,
      "recall": 8,
      "qualified": false,
      "failed_checks": [
        "cooling gate",
        "median decode clock >=1300 MHz",
        "overall clamp <=5%",
        "decode clamp <=5%",
        "load <=8"
      ],
      "exact_pair": null,
      "token_sha256": "5a0bf55103ed",
      "cool_seconds": 0,
      "runtime_id": "tensorfold0363_glm"
    },
    {
      "campaign": "glm",
      "source": "glm53_bench/summary.json",
      "status": "completed",
      "error": null,
      "id": "glm53_bench/oq2/results/run_20260930_135359/32k_retrieval_1234_draft",
      "model": "glm53_oq2",
      "mode": "MTP + context copies",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 76.55918310064327,
      "metric": "Client token count and first/last text arrival",
      "runtime_tps": 76.73836878324936,
      "ttft_seconds": 78.03262841701508,
      "gpu_mhz": 1380,
      "recall": 1,
      "qualified": false,
      "failed_checks": [
        "8/8 strict retrieval"
      ],
      "exact_pair": true,
      "token_sha256": "9e2a2160161b",
      "cool_seconds": 0,
      "runtime_id": "tensorfold0363_glm"
    },
    {
      "campaign": "glm",
      "source": "glm53_bench/summary.json",
      "status": "completed",
      "error": null,
      "id": "glm53_bench/oq2/results/run_20260930_135359/32k_retrieval_1234_serial",
      "model": "glm53_oq2",
      "mode": "Serial",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 30.075113331121354,
      "metric": "Client token count and first/last text arrival",
      "runtime_tps": 30.10199231010722,
      "ttft_seconds": 86.05913762503769,
      "gpu_mhz": 1380,
      "recall": 1,
      "qualified": false,
      "failed_checks": [
        "cooling gate",
        "8/8 strict retrieval"
      ],
      "exact_pair": null,
      "token_sha256": "9e2a2160161b",
      "cool_seconds": 0,
      "runtime_id": "tensorfold0363_glm"
    },
    {
      "campaign": "glm",
      "source": "glm53_bench/summary.json",
      "status": "completed",
      "error": null,
      "id": "glm53_bench/oq2/results/run_20260930_135359/short_code_1234_draft",
      "model": "glm53_oq2",
      "mode": "MTP + context copies",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 67.9658521661311,
      "metric": "Client token count and first/last text arrival",
      "runtime_tps": 67.96542319617416,
      "ttft_seconds": 5.364145958097652,
      "gpu_mhz": 1380,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "exact_pair": true,
      "token_sha256": "519581c1bbaf",
      "cool_seconds": 0,
      "runtime_id": "tensorfold0363_glm"
    },
    {
      "campaign": "glm",
      "source": "glm53_bench/summary.json",
      "status": "completed",
      "error": null,
      "id": "glm53_bench/oq2/results/run_20260930_135359/short_code_1234_serial",
      "model": "glm53_oq2",
      "mode": "Serial",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 30.95103479890596,
      "metric": "Client token count and first/last text arrival",
      "runtime_tps": 30.979611894961078,
      "ttft_seconds": 5.591694834060036,
      "gpu_mhz": 1380,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "exact_pair": null,
      "token_sha256": "519581c1bbaf",
      "cool_seconds": 0,
      "runtime_id": "tensorfold0363_glm"
    },
    {
      "campaign": "glm",
      "source": "glm53_bench/summary.json",
      "status": "completed",
      "error": null,
      "id": "glm53_bench/4bit/results/run_20260930_165827/256k_retrieval_1234_draft",
      "model": "glm53_4bit",
      "mode": "MTP + context copies",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 17.83134515032646,
      "metric": "Client token count and first/last text arrival",
      "runtime_tps": 17.84478470204389,
      "ttft_seconds": 1950.0563911669888,
      "gpu_mhz": 389,
      "recall": 8,
      "qualified": false,
      "failed_checks": [
        "median decode clock >=1300 MHz",
        "overall clamp <=5%",
        "decode clamp <=5%"
      ],
      "exact_pair": true,
      "token_sha256": "dedbbebd88a2",
      "cool_seconds": 0,
      "runtime_id": "tensorfold0363_glm"
    },
    {
      "campaign": "glm",
      "source": "glm53_bench/summary.json",
      "status": "completed",
      "error": null,
      "id": "glm53_bench/4bit/results/run_20260930_165827/256k_retrieval_1234_serial",
      "model": "glm53_4bit",
      "mode": "Serial",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 16.15075932335806,
      "metric": "Client token count and first/last text arrival",
      "runtime_tps": 16.163220402381427,
      "ttft_seconds": 1961.5083039579913,
      "gpu_mhz": 338,
      "recall": 8,
      "qualified": false,
      "failed_checks": [
        "cooling gate",
        "median decode clock >=1300 MHz",
        "overall clamp <=5%",
        "decode clamp <=5%",
        "load <=8"
      ],
      "exact_pair": null,
      "token_sha256": "dedbbebd88a2",
      "cool_seconds": 0,
      "runtime_id": "tensorfold0363_glm"
    },
    {
      "campaign": "glm",
      "source": "glm53_bench/summary.json",
      "status": "completed",
      "error": null,
      "id": "glm53_bench/4bit/results/run_20260930_165827/32k_retrieval_1234_draft",
      "model": "glm53_4bit",
      "mode": "MTP + context copies",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 98.9114331896185,
      "metric": "Client token count and first/last text arrival",
      "runtime_tps": 99.23103946969485,
      "ttft_seconds": 77.71721220901236,
      "gpu_mhz": 1380,
      "recall": 1,
      "qualified": false,
      "failed_checks": [
        "8/8 strict retrieval"
      ],
      "exact_pair": true,
      "token_sha256": "9e2a2160161b",
      "cool_seconds": 0,
      "runtime_id": "tensorfold0363_glm"
    },
    {
      "campaign": "glm",
      "source": "glm53_bench/summary.json",
      "status": "completed",
      "error": null,
      "id": "glm53_bench/4bit/results/run_20260930_165827/32k_retrieval_1234_serial",
      "model": "glm53_4bit",
      "mode": "Serial",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 40.694363994655866,
      "metric": "Client token count and first/last text arrival",
      "runtime_tps": 40.7484441601986,
      "ttft_seconds": 77.85766333399806,
      "gpu_mhz": 1380,
      "recall": 1,
      "qualified": false,
      "failed_checks": [
        "cooling gate",
        "8/8 strict retrieval"
      ],
      "exact_pair": null,
      "token_sha256": "9e2a2160161b",
      "cool_seconds": 0,
      "runtime_id": "tensorfold0363_glm"
    },
    {
      "campaign": "glm",
      "source": "glm53_bench/summary.json",
      "status": "completed",
      "error": null,
      "id": "glm53_bench/4bit/results/run_20260930_165827/short_code_1234_draft",
      "model": "glm53_4bit",
      "mode": "MTP + context copies",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 51.48203186136429,
      "metric": "Client token count and first/last text arrival",
      "runtime_tps": 51.57257254981243,
      "ttft_seconds": 5.792198583018035,
      "gpu_mhz": 1380,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "exact_pair": true,
      "token_sha256": "f722b61a5948",
      "cool_seconds": 0,
      "runtime_id": "tensorfold0363_glm"
    },
    {
      "campaign": "glm",
      "source": "glm53_bench/summary.json",
      "status": "completed",
      "error": null,
      "id": "glm53_bench/4bit/results/run_20260930_165827/short_code_1234_serial",
      "model": "glm53_4bit",
      "mode": "Serial",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 42.188594947452295,
      "metric": "Client token count and first/last text arrival",
      "runtime_tps": 42.24294403344106,
      "ttft_seconds": 5.779686375055462,
      "gpu_mhz": 1380,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "exact_pair": null,
      "token_sha256": "f722b61a5948",
      "cool_seconds": 0,
      "runtime_id": "tensorfold0363_glm"
    },
    {
      "campaign": "glm",
      "source": "glm53_bench/summary.json",
      "status": "completed",
      "error": null,
      "id": "glm53_bench/oq4/results/run_20260930_195227/256k_retrieval_1234_draft",
      "model": "glm53_oq4",
      "mode": "MTP + context copies",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 16.232281719275477,
      "metric": "Client token count and first/last text arrival",
      "runtime_tps": 16.24274289708789,
      "ttft_seconds": 1952.0631481669843,
      "gpu_mhz": 353.5,
      "recall": 8,
      "qualified": false,
      "failed_checks": [
        "median decode clock >=1300 MHz",
        "overall clamp <=5%",
        "decode clamp <=5%"
      ],
      "exact_pair": true,
      "token_sha256": "71e0bad36dc7",
      "cool_seconds": 0,
      "runtime_id": "tensorfold0363_glm"
    },
    {
      "campaign": "glm",
      "source": "glm53_bench/summary.json",
      "status": "completed",
      "error": null,
      "id": "glm53_bench/oq4/results/run_20260930_195227/256k_retrieval_1234_serial",
      "model": "glm53_oq4",
      "mode": "Serial",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 13.260646647733855,
      "metric": "Client token count and first/last text arrival",
      "runtime_tps": 13.267498641916216,
      "ttft_seconds": 1945.4044911658857,
      "gpu_mhz": 338,
      "recall": 8,
      "qualified": false,
      "failed_checks": [
        "median decode clock >=1300 MHz",
        "overall clamp <=5%",
        "decode clamp <=5%",
        "load <=8"
      ],
      "exact_pair": null,
      "token_sha256": "71e0bad36dc7",
      "cool_seconds": 0,
      "runtime_id": "tensorfold0363_glm"
    },
    {
      "campaign": "glm",
      "source": "glm53_bench/summary.json",
      "status": "completed",
      "error": null,
      "id": "glm53_bench/oq4/results/run_20260930_195227/32k_retrieval_1234_draft",
      "model": "glm53_oq4",
      "mode": "MTP + context copies",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 86.12173436375207,
      "metric": "Client token count and first/last text arrival",
      "runtime_tps": 86.35973032955027,
      "ttft_seconds": 77.6914454579819,
      "gpu_mhz": 1380,
      "recall": 1,
      "qualified": false,
      "failed_checks": [
        "8/8 strict retrieval"
      ],
      "exact_pair": true,
      "token_sha256": "9e2a2160161b",
      "cool_seconds": 0,
      "runtime_id": "tensorfold0363_glm"
    },
    {
      "campaign": "glm",
      "source": "glm53_bench/summary.json",
      "status": "completed",
      "error": null,
      "id": "glm53_bench/oq4/results/run_20260930_195227/32k_retrieval_1234_serial",
      "model": "glm53_oq4",
      "mode": "Serial",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 33.419286498952786,
      "metric": "Client token count and first/last text arrival",
      "runtime_tps": 33.45314082258877,
      "ttft_seconds": 80.5009511669632,
      "gpu_mhz": 1380,
      "recall": 1,
      "qualified": false,
      "failed_checks": [
        "cooling gate",
        "8/8 strict retrieval"
      ],
      "exact_pair": null,
      "token_sha256": "9e2a2160161b",
      "cool_seconds": 0,
      "runtime_id": "tensorfold0363_glm"
    },
    {
      "campaign": "glm",
      "source": "glm53_bench/summary.json",
      "status": "completed",
      "error": null,
      "id": "glm53_bench/oq4/results/run_20260930_195227/short_code_1234_draft",
      "model": "glm53_oq4",
      "mode": "MTP + context copies",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 46.78392437084825,
      "metric": "Client token count and first/last text arrival",
      "runtime_tps": 46.85392460937124,
      "ttft_seconds": 5.716455624904484,
      "gpu_mhz": 1380,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "exact_pair": true,
      "token_sha256": "f175510e6d78",
      "cool_seconds": 0,
      "runtime_id": "tensorfold0363_glm"
    },
    {
      "campaign": "glm",
      "source": "glm53_bench/summary.json",
      "status": "completed",
      "error": null,
      "id": "glm53_bench/oq4/results/run_20260930_195227/short_code_1234_serial",
      "model": "glm53_oq4",
      "mode": "Serial",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 34.200345334017136,
      "metric": "Client token count and first/last text arrival",
      "runtime_tps": 34.23649004422851,
      "ttft_seconds": 5.743218917166814,
      "gpu_mhz": 1380,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "exact_pair": null,
      "token_sha256": "f175510e6d78",
      "cool_seconds": 0,
      "runtime_id": "tensorfold0363_glm"
    },
    {
      "campaign": "flash_pilot",
      "source": "flashnext_bench/results/run_20260930_115151/summary.json",
      "status": "completed",
      "error": null,
      "id": "flashnext_bench/results/run_20260930_115151/256k_retrieval_1234_draft",
      "model": "flash4",
      "mode": "MTP + context copies",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 16.031413777406524,
      "metric": "Client token count and first/last text arrival",
      "runtime_tps": 16.041262947660336,
      "ttft_seconds": 485.9529571250314,
      "gpu_mhz": 338,
      "recall": 8,
      "qualified": false,
      "failed_checks": [
        "median decode clock >=1300 MHz",
        "overall clamp <=5%",
        "decode clamp <=5%"
      ],
      "exact_pair": true,
      "token_sha256": "f2d5e79ecd05",
      "cool_seconds": 0,
      "runtime_id": "tensorfold0363"
    },
    {
      "campaign": "flash_pilot",
      "source": "flashnext_bench/results/run_20260930_115151/summary.json",
      "status": "completed",
      "error": null,
      "id": "flashnext_bench/results/run_20260930_115151/256k_retrieval_1234_serial",
      "model": "flash4",
      "mode": "Serial",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 16.806146973948834,
      "metric": "Client token count and first/last text arrival",
      "runtime_tps": 16.818863071575297,
      "ttft_seconds": 273.774314875016,
      "gpu_mhz": 338,
      "recall": 8,
      "qualified": false,
      "failed_checks": [
        "median decode clock >=1300 MHz",
        "overall clamp <=5%",
        "decode clamp <=5%"
      ],
      "exact_pair": null,
      "token_sha256": "f2d5e79ecd05",
      "cool_seconds": 0,
      "runtime_id": "tensorfold0363"
    },
    {
      "campaign": "flash_pilot",
      "source": "flashnext_bench/results/run_20260930_115151/summary.json",
      "status": "completed",
      "error": null,
      "id": "flashnext_bench/results/run_20260930_115151/32k_retrieval_1234_draft",
      "model": "flash4",
      "mode": "MTP + context copies",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 96.51314919053293,
      "metric": "Client token count and first/last text arrival",
      "runtime_tps": 96.81953772564917,
      "ttft_seconds": 27.663858374929987,
      "gpu_mhz": 1380,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "exact_pair": true,
      "token_sha256": "aaa1b24ef225",
      "cool_seconds": 0,
      "runtime_id": "tensorfold0363"
    },
    {
      "campaign": "flash_pilot",
      "source": "flashnext_bench/results/run_20260930_115151/summary.json",
      "status": "completed",
      "error": null,
      "id": "flashnext_bench/results/run_20260930_115151/32k_retrieval_1234_serial",
      "model": "flash4",
      "mode": "Serial",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 76.52283503767123,
      "metric": "Client token count and first/last text arrival",
      "runtime_tps": 76.6928987491006,
      "ttft_seconds": 27.576390165952034,
      "gpu_mhz": 1380,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "exact_pair": null,
      "token_sha256": "aaa1b24ef225",
      "cool_seconds": 0,
      "runtime_id": "tensorfold0363"
    },
    {
      "campaign": "flash_pilot",
      "source": "flashnext_bench/results/run_20260930_115151/summary.json",
      "status": "completed",
      "error": null,
      "id": "flashnext_bench/results/run_20260930_115151/short_code_1234_draft",
      "model": "flash4",
      "mode": "MTP + context copies",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 256,
      "cached_tokens": 0,
      "decode_tps": 104.42799435110926,
      "metric": "Client token count and first/last text arrival",
      "runtime_tps": 106.07847517553644,
      "ttft_seconds": 2.4802233329974115,
      "gpu_mhz": 1377.5,
      "recall": 4,
      "qualified": false,
      "failed_checks": [
        "at least three decode clock samples",
        "8/8 strict retrieval"
      ],
      "exact_pair": true,
      "token_sha256": "e0b6e225aa45",
      "cool_seconds": 0,
      "runtime_id": "tensorfold0363"
    },
    {
      "campaign": "flash_pilot",
      "source": "flashnext_bench/results/run_20260930_115151/summary.json",
      "status": "completed",
      "error": null,
      "id": "flashnext_bench/results/run_20260930_115151/short_code_1234_serial",
      "model": "flash4",
      "mode": "Serial",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 256,
      "cached_tokens": 0,
      "decode_tps": 75.99677785127898,
      "metric": "Client token count and first/last text arrival",
      "runtime_tps": 76.72370811042622,
      "ttft_seconds": 2.583651666995138,
      "gpu_mhz": 1380,
      "recall": 4,
      "qualified": false,
      "failed_checks": [
        "8/8 strict retrieval"
      ],
      "exact_pair": null,
      "token_sha256": "e0b6e225aa45",
      "cool_seconds": 0,
      "runtime_id": "tensorfold0363"
    },
    {
      "campaign": "nemotron",
      "source": "tensorfold_bench/results/run_20260930_000006/summary.json",
      "status": "completed",
      "error": null,
      "id": "tensorfold_bench/results/run_20260930_000006/32k_retrieval_1234_draft",
      "model": "nemotron4",
      "mode": "MTP + context copies",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 160.25851491263228,
      "metric": "Client token count and first/last text arrival",
      "runtime_tps": 161.03748334604626,
      "ttft_seconds": 16.802192542003468,
      "gpu_mhz": 1380,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "exact_pair": true,
      "token_sha256": "fb15f384f22b",
      "cool_seconds": 0,
      "runtime_id": "tensorfold0363"
    },
    {
      "campaign": "nemotron",
      "source": "tensorfold_bench/results/run_20260930_000006/summary.json",
      "status": "completed",
      "error": null,
      "id": "tensorfold_bench/results/run_20260930_000006/32k_retrieval_1234_serial",
      "model": "nemotron4",
      "mode": "Serial",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 166.9323405749171,
      "metric": "Client token count and first/last text arrival",
      "runtime_tps": 167.94794697036633,
      "ttft_seconds": 14.931831834022887,
      "gpu_mhz": 1380,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "exact_pair": null,
      "token_sha256": "fb15f384f22b",
      "cool_seconds": 0,
      "runtime_id": "tensorfold0363"
    },
    {
      "campaign": "nemotron",
      "source": "tensorfold_bench/results/run_20260930_000006/summary.json",
      "status": "completed",
      "error": null,
      "id": "tensorfold_bench/results/run_20260930_000006/short_code_1234_draft",
      "model": "nemotron4",
      "mode": "MTP + context copies",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 256,
      "cached_tokens": 0,
      "decode_tps": 202.29274359366383,
      "metric": "Client token count and first/last text arrival",
      "runtime_tps": 207.17189148844045,
      "ttft_seconds": 0.9732433329336345,
      "gpu_mhz": 1331,
      "recall": 8,
      "qualified": false,
      "failed_checks": [
        "clock telemetry coverage",
        "at least three decode clock samples"
      ],
      "exact_pair": true,
      "token_sha256": "4b52b6586ad7",
      "cool_seconds": 0,
      "runtime_id": "tensorfold0363"
    },
    {
      "campaign": "nemotron",
      "source": "tensorfold_bench/results/run_20260930_000006/summary.json",
      "status": "completed",
      "error": null,
      "id": "tensorfold_bench/results/run_20260930_000006/short_code_1234_serial",
      "model": "nemotron4",
      "mode": "Serial",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 256,
      "cached_tokens": 0,
      "decode_tps": 196.85626960477845,
      "metric": "Client token count and first/last text arrival",
      "runtime_tps": 202.16998431918455,
      "ttft_seconds": 1.061883750022389,
      "gpu_mhz": 1332,
      "recall": 8,
      "qualified": false,
      "failed_checks": [
        "clock telemetry coverage",
        "at least three decode clock samples"
      ],
      "exact_pair": null,
      "token_sha256": "4b52b6586ad7",
      "cool_seconds": 0,
      "runtime_id": "tensorfold0363"
    },
    {
      "campaign": "nemotron",
      "source": "tensorfold_bench/results/run_20260930_000800/summary.json",
      "status": "completed",
      "error": null,
      "id": "tensorfold_bench/results/run_20260930_000800/256k_code_1234_draft",
      "model": "nemotron4",
      "mode": "MTP + context copies",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 8.9747076519465,
      "metric": "Client token count and first/last text arrival",
      "runtime_tps": 8.977671940803996,
      "ttft_seconds": 1090.055004834081,
      "gpu_mhz": 338,
      "recall": 7,
      "qualified": false,
      "failed_checks": [
        "cooling gate",
        "median decode clock >=1300 MHz",
        "overall clamp <=5%",
        "decode clamp <=5%",
        "fans >=3400 RPM",
        "load <=8",
        "8/8 strict retrieval"
      ],
      "exact_pair": true,
      "token_sha256": "00214fe7ca5c",
      "cool_seconds": 0,
      "runtime_id": "tensorfold0363"
    },
    {
      "campaign": "nemotron",
      "source": "tensorfold_bench/results/run_20260930_000800/summary.json",
      "status": "completed",
      "error": null,
      "id": "tensorfold_bench/results/run_20260930_000800/256k_code_1234_serial",
      "model": "nemotron4",
      "mode": "Serial",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": 260864,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 71.06741811826365,
      "metric": "Client token count and first/last text arrival",
      "runtime_tps": 71.26008542515376,
      "ttft_seconds": 310.5285092089325,
      "gpu_mhz": 942,
      "recall": 7,
      "qualified": false,
      "failed_checks": [
        "median decode clock >=1300 MHz",
        "8/8 strict retrieval"
      ],
      "exact_pair": null,
      "token_sha256": "00214fe7ca5c",
      "cool_seconds": 0,
      "runtime_id": "tensorfold0363"
    },
    {
      "campaign": "nemotron",
      "source": "tensorfold_bench/results/run_20260930_000800/summary.json",
      "status": "completed",
      "error": null,
      "id": "tensorfold_bench/results/run_20260930_000800/32k_retrieval_1234_draft",
      "model": "nemotron4",
      "mode": "MTP + context copies",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 160.11600036363308,
      "metric": "Client token count and first/last text arrival",
      "runtime_tps": 161.10777687257317,
      "ttft_seconds": 14.769094457966276,
      "gpu_mhz": 1380,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "exact_pair": true,
      "token_sha256": "fb15f384f22b",
      "cool_seconds": 0,
      "runtime_id": "tensorfold0363"
    },
    {
      "campaign": "nemotron",
      "source": "tensorfold_bench/results/run_20260930_000800/summary.json",
      "status": "completed",
      "error": null,
      "id": "tensorfold_bench/results/run_20260930_000800/32k_retrieval_1234_serial",
      "model": "nemotron4",
      "mode": "Serial",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 167.08965259566241,
      "metric": "Client token count and first/last text arrival",
      "runtime_tps": 168.05883111209192,
      "ttft_seconds": 14.939167249947786,
      "gpu_mhz": 1380,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "exact_pair": null,
      "token_sha256": "fb15f384f22b",
      "cool_seconds": 0,
      "runtime_id": "tensorfold0363"
    },
    {
      "campaign": "nemotron",
      "source": "tensorfold_bench/results/run_20260930_000800/summary.json",
      "status": "completed",
      "error": null,
      "id": "tensorfold_bench/results/run_20260930_000800/short_code_1234_draft",
      "model": "nemotron4",
      "mode": "MTP + context copies",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 256,
      "cached_tokens": 0,
      "decode_tps": 237.77059752396968,
      "metric": "Client token count and first/last text arrival",
      "runtime_tps": 244.71686468928453,
      "ttft_seconds": 0.9554892500163987,
      "gpu_mhz": null,
      "recall": 8,
      "qualified": false,
      "failed_checks": [
        "clock telemetry coverage",
        "at least three decode clock samples",
        "median decode clock >=1300 MHz",
        "overall clamp <=5%",
        "decode clamp <=5%"
      ],
      "exact_pair": true,
      "token_sha256": "4b52b6586ad7",
      "cool_seconds": 0,
      "runtime_id": "tensorfold0363"
    },
    {
      "campaign": "nemotron",
      "source": "tensorfold_bench/results/run_20260930_000800/summary.json",
      "status": "completed",
      "error": null,
      "id": "tensorfold_bench/results/run_20260930_000800/short_code_1234_serial",
      "model": "nemotron4",
      "mode": "Serial",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 256,
      "cached_tokens": 0,
      "decode_tps": 198.3184796919759,
      "metric": "Client token count and first/last text arrival",
      "runtime_tps": 203.4033501319779,
      "ttft_seconds": 0.8250487920595333,
      "gpu_mhz": 1380,
      "recall": 8,
      "qualified": false,
      "failed_checks": [
        "clock telemetry coverage",
        "at least three decode clock samples"
      ],
      "exact_pair": null,
      "token_sha256": "4b52b6586ad7",
      "cool_seconds": 0,
      "runtime_id": "tensorfold0363"
    },
    {
      "campaign": "nemotron",
      "source": "tensorfold_bench/results/run_20260930_005309/summary.json",
      "status": "incomplete",
      "error": "Fan control lost: two consecutive readings below 3400 RPM; benchmark server terminated.",
      "id": "tensorfold_bench/results/run_20260930_005309/256k_code_1234_serial",
      "model": "nemotron4",
      "mode": "Serial",
      "seed": 1234,
      "requested_input_tokens": 260864,
      "input_tokens": null,
      "output_tokens": null,
      "cached_tokens": null,
      "decode_tps": null,
      "metric": "Client token count and first/last text arrival",
      "runtime_tps": null,
      "ttft_seconds": 391.21640350006055,
      "gpu_mhz": null,
      "recall": null,
      "qualified": false,
      "failed_checks": [
        "stream_completed",
        "prompt_count_matches",
        "cold_prompt",
        "thinking_disabled",
        "draft_mode_matches",
        "draft_activity",
        "token_hash_present",
        "full_output",
        "fan_control_held",
        "median decode clock >=1300 MHz",
        "overall clamp <=5%",
        "decode clamp <=5%",
        "fans >=3400 RPM",
        "load <=8",
        "paired token-hash equality pending or failed",
        "8/8 strict retrieval"
      ],
      "exact_pair": null,
      "token_sha256": null,
      "cool_seconds": 0,
      "runtime_id": "tensorfold0363"
    },
    {
      "campaign": "nemotron",
      "source": "tensorfold_bench/results/run_20260930_005309/summary.json",
      "status": "completed",
      "error": null,
      "id": "tensorfold_bench/results/run_20260930_005309/32k_retrieval_1234_draft",
      "model": "nemotron4",
      "mode": "MTP + context copies",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 161.33578988848313,
      "metric": "Client token count and first/last text arrival",
      "runtime_tps": 162.29777371600295,
      "ttft_seconds": 14.921599374967627,
      "gpu_mhz": 1380,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "exact_pair": true,
      "token_sha256": "fb15f384f22b",
      "cool_seconds": 0,
      "runtime_id": "tensorfold0363"
    },
    {
      "campaign": "nemotron",
      "source": "tensorfold_bench/results/run_20260930_005309/summary.json",
      "status": "completed",
      "error": null,
      "id": "tensorfold_bench/results/run_20260930_005309/32k_retrieval_1234_serial",
      "model": "nemotron4",
      "mode": "Serial",
      "seed": 1234,
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 167.36178976445166,
      "metric": "Client token count and first/last text arrival",
      "runtime_tps": 168.40816014183574,
      "ttft_seconds": 14.906825290992856,
      "gpu_mhz": 1380,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "exact_pair": null,
      "token_sha256": "fb15f384f22b",
      "cool_seconds": 0,
      "runtime_id": "tensorfold0363"
    },
    {
      "campaign": "nemotron",
      "source": "tensorfold_bench/results/run_20260930_005309/summary.json",
      "status": "completed",
      "error": null,
      "id": "tensorfold_bench/results/run_20260930_005309/short_code_1234_draft",
      "model": "nemotron4",
      "mode": "MTP + context copies",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 256,
      "cached_tokens": 0,
      "decode_tps": 255.4656712785125,
      "metric": "Client token count and first/last text arrival",
      "runtime_tps": 260.60921168256294,
      "ttft_seconds": 0.9835703329881653,
      "gpu_mhz": null,
      "recall": 8,
      "qualified": false,
      "failed_checks": [
        "clock telemetry coverage",
        "at least three decode clock samples",
        "median decode clock >=1300 MHz",
        "overall clamp <=5%",
        "decode clamp <=5%"
      ],
      "exact_pair": true,
      "token_sha256": "4b52b6586ad7",
      "cool_seconds": 0,
      "runtime_id": "tensorfold0363"
    },
    {
      "campaign": "nemotron",
      "source": "tensorfold_bench/results/run_20260930_005309/summary.json",
      "status": "completed",
      "error": null,
      "id": "tensorfold_bench/results/run_20260930_005309/short_code_1234_serial",
      "model": "nemotron4",
      "mode": "Serial",
      "seed": 1234,
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 256,
      "cached_tokens": 0,
      "decode_tps": 197.9528185430607,
      "metric": "Client token count and first/last text arrival",
      "runtime_tps": 202.99666188151576,
      "ttft_seconds": 0.8295973330968991,
      "gpu_mhz": null,
      "recall": 8,
      "qualified": false,
      "failed_checks": [
        "clock telemetry coverage",
        "at least three decode clock samples",
        "median decode clock >=1300 MHz",
        "overall clamp <=5%",
        "decode clamp <=5%"
      ],
      "exact_pair": null,
      "token_sha256": "4b52b6586ad7",
      "cool_seconds": 0,
      "runtime_id": "tensorfold0363"
    },
    {
      "campaign": "e9",
      "source": "e9_bench/results/run_20260929_163156/summary.json",
      "status": "completed",
      "error": null,
      "id": "e9_bench/results/run_20260929_163156/ctx32k",
      "model": "flash_e9",
      "mode": "ENGINE native MTP",
      "requested_input_tokens": 32635,
      "input_tokens": 32635,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 75.45530346469938,
      "metric": "ENGINE generation rate, including reasoning tokens",
      "runtime_tps": 75.45530346469938,
      "client_tps": 75.58147914501043,
      "ttft_seconds": 26.37345279206056,
      "gpu_mhz": 1380,
      "recall": 8,
      "qualified": true,
      "qualification_scope": "Reported performance conditions and final-answer retrieval; no serial pair",
      "failed_checks": [],
      "exact_pair": null,
      "cool_seconds": null,
      "runtime_id": "engine_e9"
    },
    {
      "campaign": "e9",
      "source": "e9_bench/results/run_20260929_163156/summary.json",
      "status": "completed",
      "error": null,
      "id": "e9_bench/results/run_20260929_163156/ctx256k_a",
      "model": "flash_e9",
      "mode": "ENGINE native MTP",
      "requested_input_tokens": 260895,
      "input_tokens": 260895,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 24.574034102582267,
      "metric": "ENGINE generation rate, including reasoning tokens",
      "runtime_tps": 24.574034102582267,
      "client_tps": 24.815274779252302,
      "ttft_seconds": 267.41326937498525,
      "gpu_mhz": 338,
      "recall": 8,
      "qualified": false,
      "qualification_scope": "Reported performance conditions and final-answer retrieval; no serial pair",
      "failed_checks": [
        "Normal machine conditions"
      ],
      "exact_pair": null,
      "cool_seconds": null,
      "runtime_id": "engine_e9"
    },
    {
      "campaign": "e9",
      "source": "e9_bench/results/run_20260929_163156/summary.json",
      "status": "completed",
      "error": null,
      "id": "e9_bench/results/run_20260929_163156/ctx256k_b",
      "model": "flash_e9",
      "mode": "ENGINE native MTP",
      "requested_input_tokens": 260755,
      "input_tokens": 260755,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 24.253998274530883,
      "metric": "ENGINE generation rate, including reasoning tokens",
      "runtime_tps": 24.253998274530883,
      "client_tps": 24.28355143369026,
      "ttft_seconds": 684.6514497500611,
      "gpu_mhz": 338,
      "recall": 8,
      "qualified": false,
      "qualification_scope": "Reported performance conditions and final-answer retrieval; no serial pair",
      "failed_checks": [
        "Normal machine conditions"
      ],
      "exact_pair": null,
      "cool_seconds": null,
      "runtime_id": "engine_e9"
    },
    {
      "campaign": "e9",
      "source": "e9_bench/results/run_20260929_163156/summary.json",
      "status": "completed",
      "error": null,
      "id": "e9_bench/results/run_20260929_163156/ctx256k_cached",
      "model": "flash_e9",
      "mode": "ENGINE native MTP",
      "requested_input_tokens": 260895,
      "input_tokens": 260895,
      "output_tokens": 1024,
      "cached_tokens": 260894,
      "decode_tps": 59.11112589550425,
      "metric": "ENGINE generation rate, including reasoning tokens",
      "runtime_tps": 59.11112589550425,
      "client_tps": 59.233207456114464,
      "ttft_seconds": 0.4961002089548856,
      "gpu_mhz": 923,
      "recall": 8,
      "qualified": false,
      "qualification_scope": "Reported performance conditions and final-answer retrieval; no serial pair",
      "failed_checks": [
        "Normal machine conditions"
      ],
      "exact_pair": null,
      "cool_seconds": null,
      "runtime_id": "engine_e9"
    },
    {
      "campaign": "ollama",
      "source": "qwen38_27b_bench/summary.json",
      "status": "incomplete",
      "error": "KeyboardInterrupt('signal15')",
      "id": "qwen38_27b_bench/results/run_20261001_173331/256k_rep1/result.json",
      "model": "qwen27_ollama",
      "mode": "mlx_metal_v4",
      "requested_input_tokens": 260864,
      "input_tokens": null,
      "output_tokens": null,
      "cached_tokens": null,
      "decode_tps": null,
      "metric": "Client output token slope",
      "runtime_tps": null,
      "ttft_seconds": null,
      "gpu_mhz": null,
      "recall": null,
      "qualified": false,
      "failed_checks": [
        "cooling_gate",
        "decode_clock",
        "decode_clamp",
        "overall_clamp",
        "load"
      ],
      "exact_pair": null,
      "cool_seconds": 0,
      "runtime_id": "ollama_mlx_metal_v4"
    },
    {
      "campaign": "ollama",
      "source": "qwen38_27b_bench/summary.json",
      "status": "completed",
      "error": null,
      "id": "qwen38_27b_bench/results/run_20261001_173331/2k_rep1/result.json",
      "model": "qwen27_ollama",
      "mode": "mlx_metal_v4",
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 88.85896266333023,
      "metric": "Client output token slope",
      "runtime_tps": 86.5955320638279,
      "ttft_seconds": 5.275553166977305,
      "gpu_mhz": 1380,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "exact_pair": null,
      "cool_seconds": 0,
      "runtime_id": "ollama_mlx_metal_v4"
    },
    {
      "campaign": "ollama",
      "source": "qwen38_27b_bench/summary.json",
      "status": "completed",
      "error": null,
      "id": "qwen38_27b_bench/results/run_20261001_173331/2k_rep2/result.json",
      "model": "qwen27_ollama",
      "mode": "mlx_metal_v4",
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 82.23623750880385,
      "metric": "Client output token slope",
      "runtime_tps": 81.83017256467058,
      "ttft_seconds": 4.859462708901787,
      "gpu_mhz": 1380,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "exact_pair": null,
      "cool_seconds": 0,
      "runtime_id": "ollama_mlx_metal_v4"
    },
    {
      "campaign": "ollama",
      "source": "qwen38_27b_bench/summary.json",
      "status": "completed",
      "error": null,
      "id": "qwen38_27b_bench/results/run_20261001_173331/32k_rep1/result.json",
      "model": "qwen27_ollama",
      "mode": "mlx_metal_v4",
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 45.67123900289698,
      "metric": "Client output token slope",
      "runtime_tps": 46.791141179663825,
      "ttft_seconds": 86.00846358302236,
      "gpu_mhz": 1380,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "exact_pair": null,
      "cool_seconds": 0,
      "runtime_id": "ollama_mlx_metal_v4"
    },
    {
      "campaign": "ollama",
      "source": "qwen38_27b_bench/summary.json",
      "status": "completed",
      "error": null,
      "id": "qwen38_27b_bench/results/run_20261001_173331/32k_rep2/result.json",
      "model": "qwen27_ollama",
      "mode": "mlx_metal_v4",
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 48.673847783138605,
      "metric": "Client output token slope",
      "runtime_tps": 49.22088443008884,
      "ttft_seconds": 85.98376754207924,
      "gpu_mhz": 1380,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "exact_pair": null,
      "cool_seconds": 0,
      "runtime_id": "ollama_mlx_metal_v4"
    },
    {
      "campaign": "ollama",
      "source": "qwen38_27b_bench/summary.json",
      "status": "incomplete",
      "error": "Two consecutive system-load readings above16",
      "id": "qwen38_27b_bench/results/run_20261001_180533/256k_rep1/result.json",
      "model": "qwen27_ollama",
      "mode": "mlx_metal_v4",
      "requested_input_tokens": 260864,
      "input_tokens": null,
      "output_tokens": null,
      "cached_tokens": null,
      "decode_tps": null,
      "metric": "Client output token slope",
      "runtime_tps": null,
      "ttft_seconds": null,
      "gpu_mhz": null,
      "recall": null,
      "qualified": false,
      "failed_checks": [
        "decode_clock",
        "decode_clamp",
        "overall_clamp",
        "load"
      ],
      "exact_pair": null,
      "cool_seconds": 0,
      "runtime_id": "ollama_mlx_metal_v4"
    },
    {
      "campaign": "ollama",
      "source": "qwen38_27b_bench/summary.json",
      "status": "incomplete",
      "error": "Two consecutive system-load readings above16",
      "id": "qwen38_27b_bench/results/run_20261001_182755/256k_rep1/result.json",
      "model": "qwen27_ollama",
      "mode": "mlx_metal_v4",
      "requested_input_tokens": 260864,
      "input_tokens": null,
      "output_tokens": null,
      "cached_tokens": null,
      "decode_tps": null,
      "metric": "Client output token slope",
      "runtime_tps": null,
      "ttft_seconds": null,
      "gpu_mhz": null,
      "recall": null,
      "qualified": false,
      "failed_checks": [
        "decode_clock",
        "decode_clamp",
        "overall_clamp",
        "load"
      ],
      "exact_pair": null,
      "cool_seconds": 0,
      "runtime_id": "ollama_mlx_metal_v4"
    },
    {
      "campaign": "ollama",
      "source": "qwen38_27b_bench/summary.json",
      "status": "incomplete",
      "error": "Two consecutive system-load readings above16",
      "id": "qwen38_27b_bench/results/run_20261001_184030/256k_rep1/result.json",
      "model": "qwen27_ollama",
      "mode": "mlx_metal_v3",
      "requested_input_tokens": 260864,
      "input_tokens": null,
      "output_tokens": null,
      "cached_tokens": null,
      "decode_tps": null,
      "metric": "Client output token slope",
      "runtime_tps": null,
      "ttft_seconds": null,
      "gpu_mhz": null,
      "recall": null,
      "qualified": false,
      "failed_checks": [
        "decode_clock",
        "decode_clamp",
        "overall_clamp",
        "load"
      ],
      "exact_pair": null,
      "cool_seconds": 0,
      "runtime_id": "ollama_mlx_metal_v3"
    },
    {
      "campaign": "ollama",
      "source": "qwen38_27b_bench/summary.json",
      "status": "completed",
      "error": null,
      "id": "qwen38_27b_bench/results/run_20261001_184030/2k_rep1/result.json",
      "model": "qwen27_ollama",
      "mode": "mlx_metal_v3",
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 85.54525562805811,
      "metric": "Client output token slope",
      "runtime_tps": 84.37941670058717,
      "ttft_seconds": 5.40805416693823,
      "gpu_mhz": 1380,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "exact_pair": null,
      "cool_seconds": 0,
      "runtime_id": "ollama_mlx_metal_v3"
    },
    {
      "campaign": "ollama",
      "source": "qwen38_27b_bench/summary.json",
      "status": "completed",
      "error": null,
      "id": "qwen38_27b_bench/results/run_20261001_184030/2k_rep2/result.json",
      "model": "qwen27_ollama",
      "mode": "mlx_metal_v3",
      "requested_input_tokens": 2048,
      "input_tokens": 2048,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 80.68901105071672,
      "metric": "Client output token slope",
      "runtime_tps": 81.27145541124735,
      "ttft_seconds": 4.856837874977574,
      "gpu_mhz": 1380,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "exact_pair": null,
      "cool_seconds": 0,
      "runtime_id": "ollama_mlx_metal_v3"
    },
    {
      "campaign": "ollama",
      "source": "qwen38_27b_bench/summary.json",
      "status": "completed",
      "error": null,
      "id": "qwen38_27b_bench/results/run_20261001_184030/32k_rep1/result.json",
      "model": "qwen27_ollama",
      "mode": "mlx_metal_v3",
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 48.29393366432407,
      "metric": "Client output token slope",
      "runtime_tps": 48.74979585350534,
      "ttft_seconds": 85.99680870904515,
      "gpu_mhz": 1380,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "exact_pair": null,
      "cool_seconds": 0,
      "runtime_id": "ollama_mlx_metal_v3"
    },
    {
      "campaign": "ollama",
      "source": "qwen38_27b_bench/summary.json",
      "status": "completed",
      "error": null,
      "id": "qwen38_27b_bench/results/run_20261001_184030/32k_rep2/result.json",
      "model": "qwen27_ollama",
      "mode": "mlx_metal_v3",
      "requested_input_tokens": 32768,
      "input_tokens": 32768,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 48.30603115342179,
      "metric": "Client output token slope",
      "runtime_tps": 48.766102032440344,
      "ttft_seconds": 86.04983979304698,
      "gpu_mhz": 1380,
      "recall": 8,
      "qualified": true,
      "failed_checks": [],
      "exact_pair": null,
      "cool_seconds": 0,
      "runtime_id": "ollama_mlx_metal_v3"
    },
    {
      "campaign": "mlxfast",
      "source": "mlxfast_bench/results/c32k_fans/summary.json",
      "status": "completed",
      "error": null,
      "id": "mlxfast_bench/results/c32k_fans/ctl_dflash10.json",
      "model": "bonsai",
      "mode": "ctl_dflash10",
      "requested_input_tokens": 31958,
      "input_tokens": 31958,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 13.380059151847103,
      "metric": "Free-run output token slope",
      "runtime_tps": null,
      "ttft_seconds": null,
      "gpu_mhz": null,
      "recall": 5,
      "qualified": null,
      "failed_checks": [],
      "exact_pair": null,
      "cool_seconds": 0,
      "qualification_scope": "Four validated performance arms; no common qualification flag",
      "runtime_artifacts": {
        "git_head": "88cf569e885019ffe637192efcbda4aaf2c32f1c",
        "bench_worker_sha256": "4e5c6324b37d9b2a3bb256e39720a9959d6efdadbec67025f8eefde5ff57db7f",
        "metallib_sha256": "352b6986fc5e915a0e1b764a64cd26eb14822d173fe2ed01b756f2d7dd74089e"
      },
      "runtime_id": "mlxfast_control"
    },
    {
      "campaign": "mlxfast",
      "source": "mlxfast_bench/results/c32k_fans/summary.json",
      "status": "completed",
      "error": null,
      "id": "mlxfast_bench/results/c32k_fans/ctl_serial_1.json",
      "model": "bonsai",
      "mode": "ctl_serial_1",
      "requested_input_tokens": 31958,
      "input_tokens": 31958,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 27.355772767669364,
      "metric": "Free-run output token slope",
      "runtime_tps": null,
      "ttft_seconds": null,
      "gpu_mhz": null,
      "recall": 8,
      "qualified": null,
      "failed_checks": [],
      "exact_pair": null,
      "cool_seconds": 0,
      "qualification_scope": "Four validated performance arms; no common qualification flag",
      "runtime_artifacts": {
        "git_head": "88cf569e885019ffe637192efcbda4aaf2c32f1c",
        "bench_worker_sha256": "4e5c6324b37d9b2a3bb256e39720a9959d6efdadbec67025f8eefde5ff57db7f",
        "metallib_sha256": "352b6986fc5e915a0e1b764a64cd26eb14822d173fe2ed01b756f2d7dd74089e"
      },
      "runtime_id": "mlxfast_control"
    },
    {
      "campaign": "mlxfast",
      "source": "mlxfast_bench/results/c32k_fans/summary.json",
      "status": "completed",
      "error": null,
      "id": "mlxfast_bench/results/c32k_fans/winglock_dfsafe10.json",
      "model": "bonsai",
      "mode": "winglock_dfsafe10",
      "requested_input_tokens": 31958,
      "input_tokens": 31958,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 26.03134517883905,
      "metric": "Free-run output token slope",
      "runtime_tps": null,
      "ttft_seconds": null,
      "gpu_mhz": null,
      "recall": 7,
      "qualified": null,
      "failed_checks": [],
      "exact_pair": null,
      "cool_seconds": 0,
      "qualification_scope": "Four validated performance arms; no common qualification flag",
      "runtime_artifacts": {
        "git_head": "831fae740de35a106e85768d8b0534b4af422b13",
        "bench_worker_sha256": "1db295e0d6c8653b3adc1dc079d5db4b702d628b7235d7782a73a8bd367a4901",
        "metallib_sha256": "5a6ef1983bad8dbb74261524be8042fcc86d85c81c52dd5fb87b8e6f93172d7f"
      },
      "runtime_id": "mlxfast_winglock"
    },
    {
      "campaign": "mlxfast",
      "source": "mlxfast_bench/results/c32k_fans/summary.json",
      "status": "completed",
      "error": null,
      "id": "mlxfast_bench/results/c32k_fans/winglock_serial.json",
      "model": "bonsai",
      "mode": "winglock_serial",
      "requested_input_tokens": 31958,
      "input_tokens": 31958,
      "output_tokens": 1024,
      "cached_tokens": 0,
      "decode_tps": 34.06984744758276,
      "metric": "Free-run output token slope",
      "runtime_tps": null,
      "ttft_seconds": null,
      "gpu_mhz": null,
      "recall": 8,
      "qualified": null,
      "failed_checks": [],
      "exact_pair": null,
      "cool_seconds": 0,
      "qualification_scope": "Four validated performance arms; no common qualification flag",
      "runtime_artifacts": {
        "git_head": "831fae740de35a106e85768d8b0534b4af422b13",
        "bench_worker_sha256": "1db295e0d6c8653b3adc1dc079d5db4b702d628b7235d7782a73a8bd367a4901",
        "metallib_sha256": "5a6ef1983bad8dbb74261524be8042fcc86d85c81c52dd5fb87b8e6f93172d7f"
      },
      "runtime_id": "mlxfast_winglock"
    }
  ]
}