{
 "schema": 1,
 "suite": "latency-triage",
 "date": "2026-10-01T11:50:59Z",
 "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": "jeb:4b",
   "questions": "triage",
   "load_ms": 19892.5,
   "first_request_ms": 27512.8,
   "answered_by": "jeb:4b",
   "p50_ms": 7485.2,
   "p90_ms": 7652.3,
   "min_ms": 7245.8,
   "eval_p50_ms": 7478.7,
   "input_tokens": 569,
   "device": "cpu",
   "precision": "Q8_0"
  },
  {
   "model": "jeb:9b",
   "questions": "triage",
   "load_ms": 40661.3,
   "first_request_ms": 52454.1,
   "answered_by": "jeb:9b",
   "p50_ms": 11893.7,
   "p90_ms": 11951.6,
   "min_ms": 11691.7,
   "eval_p50_ms": 11887.5,
   "input_tokens": 569,
   "device": "cpu",
   "precision": "Q8_0"
  },
  {
   "model": "jeb:27b",
   "questions": "triage",
   "load_ms": 77452.5,
   "first_request_ms": 102007,
   "answered_by": "jeb:27b",
   "p50_ms": 25214.8,
   "p90_ms": 25735.5,
   "min_ms": 24664.2,
   "eval_p50_ms": 25209,
   "input_tokens": 569,
   "device": "cpu",
   "precision": "Q4_K - MEDIUM"
  },
  {
   "model": "nimble:9b",
   "questions": "triage",
   "load_ms": 73260.4,
   "first_request_ms": 124548.5,
   "answered_by": "nimble:9b",
   "p50_ms": 50730.1,
   "p90_ms": 50763.2,
   "min_ms": 50512.4,
   "eval_p50_ms": 50724.8,
   "input_tokens": 2998,
   "device": "cpu",
   "precision": "F32"
  },
  {
   "model": "jeeves:9b",
   "questions": "triage",
   "load_ms": 64561.6,
   "first_request_ms": 84633.4,
   "answered_by": "jeeves:9b",
   "p50_ms": 19521,
   "p90_ms": 19672.5,
   "min_ms": 19393,
   "eval_p50_ms": 19516.2,
   "input_tokens": 503,
   "device": "cpu",
   "precision": "F32"
  },
  {
   "model": "clm:8b",
   "questions": "triage",
   "load_ms": 62335.8,
   "first_request_ms": 74157.3,
   "answered_by": "clm:8b",
   "p50_ms": 3951.8,
   "p90_ms": 4175.1,
   "min_ms": 3774.9,
   "eval_p50_ms": 3946.2,
   "input_tokens": 342,
   "device": "cpu",
   "precision": "F32"
  },
  {
   "model": "kev:0.8b",
   "questions": "triage",
   "load_ms": 11979.8,
   "first_request_ms": 13281.1,
   "answered_by": "kev:0.8b",
   "p50_ms": 1266.6,
   "p90_ms": 1279.8,
   "min_ms": 1228.8,
   "eval_p50_ms": 1261.2,
   "input_tokens": 283,
   "device": "cpu",
   "precision": "F32"
  },
  {
   "model": "nli:deberta-v3-large",
   "questions": "triage",
   "load_ms": 7971.4,
   "first_request_ms": 8784.7,
   "answered_by": "nli:deberta-v3-large",
   "p50_ms": 790.4,
   "p90_ms": 797.1,
   "min_ms": 750.7,
   "eval_p50_ms": 784.2,
   "input_tokens": 593,
   "device": "cpu",
   "precision": "F32"
  },
  {
   "model": "nli:modernbert-large",
   "questions": "triage",
   "load_ms": 6029.9,
   "first_request_ms": 6686.9,
   "answered_by": "nli:modernbert-large",
   "p50_ms": 594.5,
   "p90_ms": 597.1,
   "min_ms": 576.5,
   "eval_p50_ms": 588.6,
   "input_tokens": 447,
   "device": "cpu",
   "precision": "F32"
  }
 ]
}
