{
  "generatedBy": "bench.mjs",
  "startedAt": "2026-08-24T06:22:17.823Z",
  "finishedAt": "2026-08-24T06:22:29.980Z",
  "machine": {
    "cpu": "Apple M1",
    "os": "Darwin 26.3",
    "arch": "arm64",
    "node": "v22.18.0",
    "python": "Python 3.13.5"
  },
  "metricCount": 99,
  "metrics": [
    {
      "metric": "experiment_scope",
      "value": "same model, quantization, hardware, chat prompt, seed, context, and output cap"
    },
    {
      "metric": "model_name",
      "value": "qwen2.5:1.5b"
    },
    {
      "metric": "model_quantization",
      "value": "Q4_K_M"
    },
    {
      "metric": "model_path_sha256",
      "value": "183715c435899236895da3869489cc30ac241476b4971a20285b1a462818a5b4"
    },
    {
      "metric": "model_file_size_mb",
      "value": 940.37,
      "unit": "MB"
    },
    {
      "metric": "workload_prompt_sha256",
      "value": "8dede1a3a7366e27d5fe8acb75ede788c9f4e80226c8c4f6016788e483d530f4"
    },
    {
      "metric": "context_tokens",
      "value": 4096,
      "unit": "tokens"
    },
    {
      "metric": "max_output_tokens",
      "value": 128,
      "unit": "tokens"
    },
    {
      "metric": "temperature",
      "value": 0
    },
    {
      "metric": "seed",
      "value": 42
    },
    {
      "metric": "repetitions_per_surface",
      "value": 3,
      "unit": "runs"
    },
    {
      "metric": "hardware",
      "value": "arm64 / Darwin 26.3"
    },
    {
      "metric": "ollama_version",
      "value": "0.32.13"
    },
    {
      "metric": "llama_cpp_version",
      "value": "version: 1 (0b1bad14f)"
    },
    {
      "metric": "ollama_server_health_ms",
      "value": 155.2,
      "unit": "ms",
      "note": "fresh Ollama process start to /api/tags"
    },
    {
      "metric": "ollama_run_1_wall_ms",
      "value": 2720.94,
      "unit": "ms",
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "ollama_run_1_success",
      "value": 1,
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "ollama_run_1_output_valid",
      "value": 1,
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "ollama_run_1_prompt_tokens",
      "value": 54,
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "ollama_run_1_output_tokens",
      "value": 53,
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "ollama_run_1_tokens_per_s",
      "value": 48.078,
      "unit": "tok/s",
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "ollama_run_1_prompt_tokens_per_s",
      "value": 388.492,
      "unit": "tok/s",
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "ollama_run_1_load_ms",
      "value": 1463.05,
      "unit": "ms",
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "ollama_run_1_rss_mb",
      "value": 1164.42,
      "unit": "MB",
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "ollama_run_1_output_sha256",
      "value": "4cc0715ab4581f548e1646d217bff03f89e66bd2d9ceb69614a81e17924cbe88",
      "note": "hash of returned assistant text"
    },
    {
      "metric": "ollama_run_2_wall_ms",
      "value": 1268.53,
      "unit": "ms",
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "ollama_run_2_success",
      "value": 1,
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "ollama_run_2_output_valid",
      "value": 1,
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "ollama_run_2_prompt_tokens",
      "value": 54,
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "ollama_run_2_output_tokens",
      "value": 53,
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "ollama_run_2_tokens_per_s",
      "value": 47.64,
      "unit": "tok/s",
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "ollama_run_2_prompt_tokens_per_s",
      "value": 2568.493,
      "unit": "tok/s",
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "ollama_run_2_load_ms",
      "value": 131.67,
      "unit": "ms",
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "ollama_run_2_rss_mb",
      "value": 1167.03,
      "unit": "MB",
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "ollama_run_2_output_sha256",
      "value": "4cc0715ab4581f548e1646d217bff03f89e66bd2d9ceb69614a81e17924cbe88",
      "note": "hash of returned assistant text"
    },
    {
      "metric": "ollama_run_3_wall_ms",
      "value": 1246.03,
      "unit": "ms",
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "ollama_run_3_success",
      "value": 1,
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "ollama_run_3_output_valid",
      "value": 1,
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "ollama_run_3_prompt_tokens",
      "value": 54,
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "ollama_run_3_output_tokens",
      "value": 53,
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "ollama_run_3_tokens_per_s",
      "value": 48.448,
      "unit": "tok/s",
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "ollama_run_3_prompt_tokens_per_s",
      "value": 2589.68,
      "unit": "tok/s",
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "ollama_run_3_load_ms",
      "value": 126.89,
      "unit": "ms",
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "ollama_run_3_rss_mb",
      "value": 1132.52,
      "unit": "MB",
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "ollama_run_3_output_sha256",
      "value": "4cc0715ab4581f548e1646d217bff03f89e66bd2d9ceb69614a81e17924cbe88",
      "note": "hash of returned assistant text"
    },
    {
      "metric": "ollama_median_wall_ms",
      "value": 1268.53,
      "unit": "ms",
      "n": 3,
      "note": "median of 3 repetitions"
    },
    {
      "metric": "ollama_median_success",
      "value": 1,
      "n": 3,
      "note": "median of 3 repetitions"
    },
    {
      "metric": "ollama_median_output_valid",
      "value": 1,
      "n": 3,
      "note": "median of 3 repetitions"
    },
    {
      "metric": "ollama_median_prompt_tokens",
      "value": 54,
      "n": 3,
      "note": "median of 3 repetitions"
    },
    {
      "metric": "ollama_median_output_tokens",
      "value": 53,
      "n": 3,
      "note": "median of 3 repetitions"
    },
    {
      "metric": "ollama_median_tokens_per_s",
      "value": 48.078,
      "unit": "tok/s",
      "n": 3,
      "note": "median of 3 repetitions"
    },
    {
      "metric": "ollama_median_prompt_tokens_per_s",
      "value": 2568.493,
      "unit": "tok/s",
      "n": 3,
      "note": "median of 3 repetitions"
    },
    {
      "metric": "ollama_median_load_ms",
      "value": 131.67,
      "unit": "ms",
      "n": 3,
      "note": "median of 3 repetitions"
    },
    {
      "metric": "ollama_median_rss_mb",
      "value": 1164.42,
      "unit": "MB",
      "n": 3,
      "note": "median of 3 repetitions"
    },
    {
      "metric": "ollama_log_path",
      "value": "/tmp/agenticwire-ollama-vs-llama-cpp-ollama.log"
    },
    {
      "metric": "llama_cpp_server_health_ms",
      "value": 768.05,
      "unit": "ms",
      "note": "fresh llama-server start to /health; includes model load"
    },
    {
      "metric": "llama_cpp_run_1_wall_ms",
      "value": 1268.76,
      "unit": "ms",
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "llama_cpp_run_1_success",
      "value": 1,
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "llama_cpp_run_1_output_valid",
      "value": 1,
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "llama_cpp_run_1_prompt_tokens",
      "value": 54,
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "llama_cpp_run_1_output_tokens",
      "value": 53,
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "llama_cpp_run_1_tokens_per_s",
      "value": 47.17,
      "unit": "tok/s",
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "llama_cpp_run_1_prompt_tokens_per_s",
      "value": 404.367,
      "unit": "tok/s",
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "llama_cpp_run_1_load_ms",
      "value": 0,
      "unit": "ms",
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "llama_cpp_run_1_rss_mb",
      "value": 1130.14,
      "unit": "MB",
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "llama_cpp_run_1_output_sha256",
      "value": "4cc0715ab4581f548e1646d217bff03f89e66bd2d9ceb69614a81e17924cbe88",
      "note": "hash of returned assistant text"
    },
    {
      "metric": "llama_cpp_run_2_wall_ms",
      "value": 1273.13,
      "unit": "ms",
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "llama_cpp_run_2_success",
      "value": 1,
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "llama_cpp_run_2_output_valid",
      "value": 1,
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "llama_cpp_run_2_prompt_tokens",
      "value": 54,
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "llama_cpp_run_2_output_tokens",
      "value": 53,
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "llama_cpp_run_2_tokens_per_s",
      "value": 46.662,
      "unit": "tok/s",
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "llama_cpp_run_2_prompt_tokens_per_s",
      "value": 409.864,
      "unit": "tok/s",
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "llama_cpp_run_2_load_ms",
      "value": 0,
      "unit": "ms",
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "llama_cpp_run_2_rss_mb",
      "value": 1110.23,
      "unit": "MB",
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "llama_cpp_run_2_output_sha256",
      "value": "4cc0715ab4581f548e1646d217bff03f89e66bd2d9ceb69614a81e17924cbe88",
      "note": "hash of returned assistant text"
    },
    {
      "metric": "llama_cpp_run_3_wall_ms",
      "value": 1251.14,
      "unit": "ms",
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "llama_cpp_run_3_success",
      "value": 1,
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "llama_cpp_run_3_output_valid",
      "value": 1,
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "llama_cpp_run_3_prompt_tokens",
      "value": 54,
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "llama_cpp_run_3_output_tokens",
      "value": 53,
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "llama_cpp_run_3_tokens_per_s",
      "value": 47.183,
      "unit": "tok/s",
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "llama_cpp_run_3_prompt_tokens_per_s",
      "value": 439.686,
      "unit": "tok/s",
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "llama_cpp_run_3_load_ms",
      "value": 0,
      "unit": "ms",
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "llama_cpp_run_3_rss_mb",
      "value": 1110.02,
      "unit": "MB",
      "note": "one same-prompt local inference run"
    },
    {
      "metric": "llama_cpp_run_3_output_sha256",
      "value": "4cc0715ab4581f548e1646d217bff03f89e66bd2d9ceb69614a81e17924cbe88",
      "note": "hash of returned assistant text"
    },
    {
      "metric": "llama_cpp_median_wall_ms",
      "value": 1268.76,
      "unit": "ms",
      "n": 3,
      "note": "median of 3 repetitions"
    },
    {
      "metric": "llama_cpp_median_success",
      "value": 1,
      "n": 3,
      "note": "median of 3 repetitions"
    },
    {
      "metric": "llama_cpp_median_output_valid",
      "value": 1,
      "n": 3,
      "note": "median of 3 repetitions"
    },
    {
      "metric": "llama_cpp_median_prompt_tokens",
      "value": 54,
      "n": 3,
      "note": "median of 3 repetitions"
    },
    {
      "metric": "llama_cpp_median_output_tokens",
      "value": 53,
      "n": 3,
      "note": "median of 3 repetitions"
    },
    {
      "metric": "llama_cpp_median_tokens_per_s",
      "value": 47.17,
      "unit": "tok/s",
      "n": 3,
      "note": "median of 3 repetitions"
    },
    {
      "metric": "llama_cpp_median_prompt_tokens_per_s",
      "value": 409.864,
      "unit": "tok/s",
      "n": 3,
      "note": "median of 3 repetitions"
    },
    {
      "metric": "llama_cpp_median_load_ms",
      "value": 0,
      "unit": "ms",
      "n": 3,
      "note": "median of 3 repetitions"
    },
    {
      "metric": "llama_cpp_median_rss_mb",
      "value": 1110.23,
      "unit": "MB",
      "n": 3,
      "note": "median of 3 repetitions"
    },
    {
      "metric": "llama_cpp_log_path",
      "value": "/tmp/agenticwire-ollama-vs-llama-cpp-llama.log"
    },
    {
      "metric": "same_output_hash",
      "value": 1,
      "note": "same deterministic chat output hash across surfaces"
    },
    {
      "metric": "all_successful_runs",
      "value": 6,
      "unit": "runs"
    },
    {
      "metric": "all_valid_outputs",
      "value": 6,
      "unit": "runs"
    }
  ],
  "log": [
    ""
  ]
}