{
  "bench_run_identity": {
    "api_mode": "chat",
    "base_url": "http://127.0.0.1:18260",
    "created_at_utc": "2026-06-23T06:41:04.015992+00:00",
    "max_tokens": 512,
    "model": "gemma4-26b-a4b-q8",
    "prompt_mode": "long",
    "prompt_tokens_requested": 512,
    "repeats": 8,
    "seed": 1,
    "usage_required": true
  },
  "bench_summary": {
    "completion_tokens_total": 4096,
    "elapsed_s": {
      "cv": 0.0011246973614698172,
      "max": 12.415953420015285,
      "mean": 12.381748673375114,
      "median": 12.37641503699706,
      "min": 12.375048496993259,
      "stdev": 0.013925720063327401
    },
    "requests": 8,
    "tok_s_after_ttft": {
      "cv": 0.0001995828023148337,
      "max": 42.72772797604998,
      "mean": 42.716267799445156,
      "median": 42.71819081833714,
      "min": 42.70265770141791,
      "stdev": 0.008525432431844159
    },
    "tok_s_wall": {
      "cv": 0.0011220984953366035,
      "max": 41.37357523280815,
      "mean": 41.351232276832015,
      "median": 41.36900702218013,
      "min": 41.23726810819251,
      "stdev": 0.0464001555181476
    },
    "ttft_s": {
      "cv": 0.03685287836472445,
      "max": 0.43151944200508296,
      "mean": 0.39568237537605455,
      "median": 0.3905731765116798,
      "min": 0.38749196100980043,
      "stdev": 0.014582034450798979
    }
  },
  "canary_pass_all": true,
  "canary_rows_completed": 384,
  "label": "gemma4-q8-gpu0-currentbest-longprompt-deep-20260623T0945",
  "launcher_identity": {
    "batch_size": "512",
    "cache_type_k": "f16",
    "cache_type_v": "f16",
    "ctx_size": "8192",
    "extra_llama_args": "--parallel 1 --cache-ram 0",
    "flash_attn": "off",
    "ggml_sycl_disable_dnn": null,
    "ggml_sycl_disable_graph": null,
    "ggml_sycl_disable_opt": "0",
    "gpu_index": "0",
    "oneapi_device_selector": null,
    "poll": "50",
    "port": "18260",
    "reasoning": "off",
    "threads": "16",
    "ubatch_size": "64"
  },
  "model_file_bytes": 27636230944,
  "model_path": "/mnt/fast-ai/llm-models/gemma4-26b-a4b-it-q8-gguf/gemma-4-26B-A4B-it-UD-Q8_K_XL.gguf",
  "run_dir": "/home/steve/qwen36-results-main/data/gemma4-q8-gpu0-currentbest-longprompt-deep-20260623T0945",
  "server_log": "/mnt/fast-ai/bench-results/gemma4-26b-a4b-q8/servers/gemma4-q8-gpu0-currentbest-longprompt-deep-20260623T0945.server.log"
}
