{
  "schema": 1,
  "benchmark": "hermes-tool-prefix-cache",
  "date": "2026-07-28",
  "server_commit": "c3b71e4b2ed2d08db96d55e7985f20035aa08d6f",
  "hardware": "AMD Ryzen AI MAX+ 395 with Radeon 8060S (gfx1151)",
  "runtime": "ROCm 7.2.4",
  "model": "Qwen3.6-27B Q4_K_M",
  "client": "Hermes Agent 0.19.0",
  "settings": {
    "max_context_tokens": 65536,
    "max_output_tokens": 256,
    "temperature": 0,
    "draft_model": null,
    "prefill_cache_slots": 2,
    "enabled_prefix_cache_slots": 4,
    "tools": 18
  },
  "task": {
    "commands": [
      "git rev-parse --short HEAD",
      "wc -l README.md",
      "grep -c 'Qwen3.6' server/README.md"
    ],
    "expected_values": ["c3b71e4", "493 README.md", "24"],
    "cache_disabled_correct": true,
    "cache_enabled_correct": true
  },
  "cache_disabled": {
    "wall_seconds": 243.48,
    "total_prefill_seconds": 221.4,
    "warm_prefill_seconds": 167.5,
    "requests": [
      {"prompt_tokens": 15615, "output_tokens": 31, "prefill_seconds": 53.9, "decode_seconds": 3.0, "cache_hit": false, "cached_prefix_tokens": 0},
      {"prompt_tokens": 15680, "output_tokens": 28, "prefill_seconds": 55.1, "decode_seconds": 2.7, "cache_hit": false, "cached_prefix_tokens": 0},
      {"prompt_tokens": 15742, "output_tokens": 39, "prefill_seconds": 55.8, "decode_seconds": 3.8, "cache_hit": false, "cached_prefix_tokens": 0},
      {"prompt_tokens": 15810, "output_tokens": 81, "prefill_seconds": 56.6, "decode_seconds": 8.1, "cache_hit": false, "cached_prefix_tokens": 0}
    ]
  },
  "cache_enabled": {
    "wall_seconds": 82.45,
    "total_prefill_seconds": 60.8,
    "warm_prefill_seconds": 5.2,
    "requests": [
      {"prompt_tokens": 15615, "output_tokens": 31, "prefill_seconds": 55.6, "decode_seconds": 3.1, "cache_hit": false, "cached_prefix_tokens": 0},
      {"prompt_tokens": 15680, "output_tokens": 28, "prefill_seconds": 1.4, "decode_seconds": 2.8, "cache_hit": true, "cached_prefix_tokens": 15360},
      {"prompt_tokens": 15742, "output_tokens": 39, "prefill_seconds": 1.7, "decode_seconds": 3.9, "cache_hit": true, "cached_prefix_tokens": 15360},
      {"prompt_tokens": 15810, "output_tokens": 69, "prefill_seconds": 2.1, "decode_seconds": 7.0, "cache_hit": true, "cached_prefix_tokens": 15360}
    ]
  },
  "speedup": {
    "whole_task": 2.953,
    "total_prefill": 3.641,
    "warm_prefill": 32.212
  }
}