{
 "schema": 1,
 "suite": "latency-triage",
 "date": "2026-10-01T11:26:25Z",
 "ollaya": "0.8.0",
 "machine": "rtx-4090",
 "machine_detail": {
  "id": "rtx-4090",
  "cpu": "13th Gen Intel(R) Core(TM) i9-13900K",
  "threads": 24,
  "gpus": [
   {
    "name": "NVIDIA GeForce RTX 4090",
    "driver": "616.92",
    "memory": "24564 MiB"
   }
  ],
  "os": "Linux 6.18.33.2-microsoft-standard-WSL2"
 },
 "device": "cpu",
 "protocol": {
  "preset": "triage",
  "questions": 5,
  "states": 30,
  "warm": 2,
  "n": 10,
  "client": "one request at a time over HTTP to /api/decide, timed by the client; each request a different short message, in turn"
 },
 "note": "",
 "results": [
  {
   "model": "cygnet:12b",
   "questions": "triage",
   "load_ms": 14439,
   "first_request_ms": 37657.5,
   "answered_by": "cygnet:12b",
   "p50_ms": 23582.8,
   "p90_ms": 24231.8,
   "min_ms": 23150.1,
   "eval_p50_ms": 23581.9,
   "input_tokens": 746,
   "device": "cpu",
   "precision": "Q8_0"
  },
  {
   "model": "decider:0.8b",
   "questions": "triage",
   "load_ms": 5367.6,
   "first_request_ms": 7086.4,
   "answered_by": "decider:0.8b",
   "p50_ms": 1810.7,
   "p90_ms": 2010.5,
   "min_ms": 1788.6,
   "eval_p50_ms": 1809.9,
   "input_tokens": 552,
   "device": "cpu",
   "precision": "F32"
  },
  {
   "model": "decider:2b",
   "questions": "triage",
   "load_ms": 20804.9,
   "first_request_ms": 24225.3,
   "answered_by": "decider:2b",
   "p50_ms": 3357,
   "p90_ms": 3366.3,
   "min_ms": 3340.6,
   "eval_p50_ms": 3356.3,
   "input_tokens": 552,
   "device": "cpu",
   "precision": "F32"
  },
  {
   "model": "decider:2b-vision",
   "questions": "triage",
   "load_ms": 16991,
   "first_request_ms": 19303.5,
   "answered_by": "decider:2b-vision",
   "p50_ms": 2014.8,
   "p90_ms": 2261.6,
   "min_ms": 1972.1,
   "eval_p50_ms": 2014.2,
   "input_tokens": 381,
   "device": "cpu",
   "precision": "F32"
  },
  {
   "model": "decider:4b",
   "questions": "triage",
   "load_ms": 5487.2,
   "first_request_ms": 14497.5,
   "answered_by": "decider:4b",
   "p50_ms": 8778.6,
   "p90_ms": 8851.5,
   "min_ms": 8731,
   "eval_p50_ms": 8778,
   "input_tokens": 552,
   "device": "cpu",
   "precision": "F32"
  },
  {
   "model": "decision:eos",
   "questions": "triage",
   "load_ms": 5096.1,
   "first_request_ms": 7178.8,
   "answered_by": "decision:eos",
   "p50_ms": 2021.9,
   "p90_ms": 2048.8,
   "min_ms": 2002.8,
   "eval_p50_ms": 2021.3,
   "input_tokens": 685,
   "device": "cpu",
   "precision": "F32"
  },
  {
   "model": "kev:4b",
   "questions": "triage",
   "load_ms": 5436.3,
   "first_request_ms": 11700.2,
   "answered_by": "kev:4b",
   "p50_ms": 6080.9,
   "p90_ms": 6240.7,
   "min_ms": 5999.5,
   "eval_p50_ms": 6080.3,
   "input_tokens": 283,
   "device": "cpu",
   "precision": "F32"
  },
  {
   "model": "kev:9b",
   "questions": "triage",
   "load_ms": 9043.6,
   "first_request_ms": 19765,
   "answered_by": "kev:9b",
   "p50_ms": 10474.7,
   "p90_ms": 10490.6,
   "min_ms": 10429.7,
   "eval_p50_ms": 10474,
   "input_tokens": 288,
   "device": "cpu",
   "precision": "F32"
  },
  {
   "model": "laya:en",
   "questions": "triage",
   "load_ms": 2507.1,
   "first_request_ms": 3156.9,
   "answered_by": "laya:en",
   "p50_ms": 619.1,
   "p90_ms": 625.2,
   "min_ms": 608.4,
   "eval_p50_ms": 618.5,
   "input_tokens": 364,
   "device": "cpu",
   "precision": "F32"
  },
  {
   "model": "laya:multilingual",
   "questions": "triage",
   "load_ms": 1432.8,
   "first_request_ms": 1748.2,
   "answered_by": "laya:multilingual",
   "p50_ms": 307.5,
   "p90_ms": 311.9,
   "min_ms": 301.2,
   "eval_p50_ms": 306.9,
   "input_tokens": 371,
   "device": "cpu",
   "precision": "F32"
  },
  {
   "model": "winnow:12b",
   "questions": "triage",
   "load_ms": 10234.2,
   "first_request_ms": 25382.3,
   "answered_by": "winnow:12b",
   "p50_ms": 14824.4,
   "p90_ms": 15114.7,
   "min_ms": 14728.1,
   "eval_p50_ms": 14823.6,
   "input_tokens": 706,
   "device": "cpu",
   "precision": "Q8_0"
  },
  {
   "model": "winnow:e4b",
   "questions": "triage",
   "load_ms": 4464.4,
   "first_request_ms": 10558.7,
   "answered_by": "winnow:e4b",
   "p50_ms": 6119,
   "p90_ms": 6192.1,
   "min_ms": 6097.9,
   "eval_p50_ms": 6118.3,
   "input_tokens": 686,
   "device": "cpu",
   "precision": "Q8_0"
  },
  {
   "model": "gliclass:latest",
   "questions": "triage",
   "load_ms": 2650.9,
   "first_request_ms": 3252.4,
   "answered_by": "gliclass:latest",
   "p50_ms": 563,
   "p90_ms": 572.1,
   "min_ms": 547.3,
   "eval_p50_ms": 562.5,
   "input_tokens": 283,
   "device": "cpu",
   "precision": "F32"
  },
  {
   "model": "jevk5:latest",
   "questions": "triage",
   "load_ms": 4219,
   "first_request_ms": 13627.5,
   "answered_by": "jevk5:latest",
   "p50_ms": 9466.2,
   "p90_ms": 9583.3,
   "min_ms": 9262.5,
   "eval_p50_ms": 9465.4,
   "input_tokens": 747,
   "device": "cpu",
   "precision": "Q8_0"
  },
  {
   "model": "nli:latest",
   "questions": "triage",
   "load_ms": 3047.5,
   "first_request_ms": 3840,
   "answered_by": "nli:latest",
   "p50_ms": 781.3,
   "p90_ms": 792.7,
   "min_ms": 743.4,
   "eval_p50_ms": 780.7,
   "input_tokens": 606,
   "device": "cpu",
   "precision": "F32"
  },
  {
   "model": "qwen3guard:latest",
   "questions": "builtin (4)",
   "load_ms": 0,
   "first_request_ms": 793.7,
   "answered_by": "qwen3guard:latest",
   "p50_ms": 801.5,
   "p90_ms": 811.4,
   "min_ms": 790.4,
   "eval_p50_ms": 800.9,
   "input_tokens": 642,
   "device": "cpu",
   "precision": "F32"
  },
  {
   "model": "von:latest",
   "questions": "triage",
   "load_ms": 1888.7,
   "first_request_ms": 2668.1,
   "answered_by": "von:latest",
   "p50_ms": 753.1,
   "p90_ms": 777.6,
   "min_ms": 733.3,
   "eval_p50_ms": 752.5,
   "input_tokens": 374,
   "device": "cpu",
   "precision": "F32"
  }
 ]
}
