joule-bench/results_mix.json
David Charlot eda1628ec3 joule-bench: Resolution x energy/cost coding-agent benchmark
A solver-agnostic harness that grades coding-agent task success with an
executable test oracle (tests fail before, pass after; pristine test files
restored to prevent gaming) and pairs it with measured energy (jouleclaw
IOReport, idle-subtracted, sub-window readings flagged unreliable), vendor-billed
dollars, and token-estimated cloud energy. Solvers: reference, noop, claude,
gemini, and cascade (the real joule_core engine via joule-coded). --passes shows
L0 cache amortization.

Headline (RESULTS.md): on a realistic operation mix, the cascade and frontier
claude both resolve 100%, but the cascade costs ~10x less per task -- mechanical
ops close locally at L1 for free, novel ops escalate to the same model once, and
repeats are free from the L0 cache.

Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
2026-06-22 07:18:36 -04:00

378 lines
No EOL
8.9 KiB
JSON

{
"energy_meter": "Measured (ModelBased, jouleclaw host counter, idle≈134.9 W subtracted)",
"frontier": [
{
"cloud_joules_per_resolved": 0.0,
"cost_per_resolved": 0.020660274999999995,
"joules_per_resolved": 0.0,
"median_wall_ms": 1298,
"provenance": "ModelBased",
"reliable_energy_runs": 10,
"resolution_pct": 100.0,
"resolved": 10,
"solver": "cascade",
"tasks": 10,
"total_cloud_joules": 0.0,
"total_cost_usd": 0.20660274999999995,
"total_joules": 0.0
},
{
"cloud_joules_per_resolved": 29302.379999999997,
"cost_per_resolved": 0.20433379999999995,
"joules_per_resolved": 0.0,
"median_wall_ms": 22131,
"provenance": "ModelBased",
"reliable_energy_runs": 10,
"resolution_pct": 100.0,
"resolved": 10,
"solver": "claude",
"tasks": 10,
"total_cloud_joules": 293023.8,
"total_cost_usd": 2.0433379999999994,
"total_joules": 0.0
}
],
"runs": [
{
"baseline_failed": true,
"cost_usd": 0.0,
"energy_reliable": true,
"est_cloud_uj": 0,
"input_tokens": null,
"microjoules": 0,
"note": "",
"output_tokens": null,
"pass": 1,
"provenance": "ModelBased",
"resolved": true,
"solver": "cascade",
"solver_note": "L1",
"task": "const",
"wall_ms": 1313
},
{
"baseline_failed": true,
"cost_usd": 0.0,
"energy_reliable": true,
"est_cloud_uj": 0,
"input_tokens": null,
"microjoules": 0,
"note": "",
"output_tokens": null,
"pass": 1,
"provenance": "ModelBased",
"resolved": true,
"solver": "cascade",
"solver_note": "L1",
"task": "greeting",
"wall_ms": 1298
},
{
"baseline_failed": true,
"cost_usd": 0.20660274999999995,
"energy_reliable": true,
"est_cloud_uj": 0,
"input_tokens": null,
"microjoules": 0,
"note": "",
"output_tokens": null,
"pass": 1,
"provenance": "ModelBased",
"resolved": true,
"solver": "cascade",
"solver_note": "L4",
"task": "logic",
"wall_ms": 23145
},
{
"baseline_failed": true,
"cost_usd": 0.0,
"energy_reliable": true,
"est_cloud_uj": 0,
"input_tokens": null,
"microjoules": 0,
"note": "",
"output_tokens": null,
"pass": 1,
"provenance": "ModelBased",
"resolved": true,
"solver": "cascade",
"solver_note": "L1",
"task": "rename",
"wall_ms": 1778
},
{
"baseline_failed": true,
"cost_usd": 0.0,
"energy_reliable": true,
"est_cloud_uj": 0,
"input_tokens": null,
"microjoules": 0,
"note": "",
"output_tokens": null,
"pass": 1,
"provenance": "ModelBased",
"resolved": true,
"solver": "cascade",
"solver_note": "L1",
"task": "typo",
"wall_ms": 1778
},
{
"baseline_failed": true,
"cost_usd": 0.188752,
"energy_reliable": true,
"est_cloud_uj": 24369800000,
"input_tokens": 116399,
"microjoules": 0,
"note": "",
"output_tokens": 545,
"pass": 1,
"provenance": "ModelBased",
"resolved": true,
"solver": "claude",
"solver_note": null,
"task": "const",
"wall_ms": 16374
},
{
"baseline_failed": true,
"cost_usd": 0.1864925,
"energy_reliable": true,
"est_cloud_uj": 24188800000,
"input_tokens": 115974,
"microjoules": 0,
"note": "",
"output_tokens": 497,
"pass": 1,
"provenance": "ModelBased",
"resolved": true,
"solver": "claude",
"solver_note": null,
"task": "greeting",
"wall_ms": 16516
},
{
"baseline_failed": true,
"cost_usd": 0.213112,
"energy_reliable": true,
"est_cloud_uj": 31848200000,
"input_tokens": 152301,
"microjoules": 0,
"note": "",
"output_tokens": 694,
"pass": 1,
"provenance": "ModelBased",
"resolved": true,
"solver": "claude",
"solver_note": null,
"task": "logic",
"wall_ms": 222593
},
{
"baseline_failed": true,
"cost_usd": 0.215365,
"energy_reliable": true,
"est_cloud_uj": 32044000000,
"input_tokens": 152610,
"microjoules": 0,
"note": "",
"output_tokens": 761,
"pass": 1,
"provenance": "ModelBased",
"resolved": true,
"solver": "claude",
"solver_note": null,
"task": "rename",
"wall_ms": 20984
},
{
"baseline_failed": true,
"cost_usd": 0.2314895,
"energy_reliable": true,
"est_cloud_uj": 38248800000,
"input_tokens": 182804,
"microjoules": 0,
"note": "",
"output_tokens": 844,
"pass": 1,
"provenance": "ModelBased",
"resolved": true,
"solver": "claude",
"solver_note": null,
"task": "typo",
"wall_ms": 23024
},
{
"baseline_failed": true,
"cost_usd": 0.0,
"energy_reliable": true,
"est_cloud_uj": 0,
"input_tokens": null,
"microjoules": 0,
"note": "",
"output_tokens": null,
"pass": 2,
"provenance": "ModelBased",
"resolved": true,
"solver": "cascade",
"solver_note": "L0",
"task": "const",
"wall_ms": 167
},
{
"baseline_failed": true,
"cost_usd": 0.0,
"energy_reliable": true,
"est_cloud_uj": 0,
"input_tokens": null,
"microjoules": 0,
"note": "",
"output_tokens": null,
"pass": 2,
"provenance": "ModelBased",
"resolved": true,
"solver": "cascade",
"solver_note": "L0",
"task": "greeting",
"wall_ms": 166
},
{
"baseline_failed": true,
"cost_usd": 0.0,
"energy_reliable": true,
"est_cloud_uj": 0,
"input_tokens": null,
"microjoules": 0,
"note": "",
"output_tokens": null,
"pass": 2,
"provenance": "ModelBased",
"resolved": true,
"solver": "cascade",
"solver_note": "L0",
"task": "logic",
"wall_ms": 165
},
{
"baseline_failed": true,
"cost_usd": 0.0,
"energy_reliable": true,
"est_cloud_uj": 0,
"input_tokens": null,
"microjoules": 0,
"note": "",
"output_tokens": null,
"pass": 2,
"provenance": "ModelBased",
"resolved": true,
"solver": "cascade",
"solver_note": "L0",
"task": "rename",
"wall_ms": 168
},
{
"baseline_failed": true,
"cost_usd": 0.0,
"energy_reliable": true,
"est_cloud_uj": 0,
"input_tokens": null,
"microjoules": 0,
"note": "",
"output_tokens": null,
"pass": 2,
"provenance": "ModelBased",
"resolved": true,
"solver": "cascade",
"solver_note": "L0",
"task": "typo",
"wall_ms": 170
},
{
"baseline_failed": true,
"cost_usd": 0.18969975,
"energy_reliable": true,
"est_cloud_uj": 24445000000,
"input_tokens": 116495,
"microjoules": 0,
"note": "",
"output_tokens": 573,
"pass": 2,
"provenance": "ModelBased",
"resolved": true,
"solver": "claude",
"solver_note": null,
"task": "const",
"wall_ms": 23325
},
{
"baseline_failed": true,
"cost_usd": 0.185425,
"energy_reliable": true,
"est_cloud_uj": 24103000000,
"input_tokens": 115845,
"microjoules": 0,
"note": "",
"output_tokens": 467,
"pass": 2,
"provenance": "ModelBased",
"resolved": true,
"solver": "claude",
"solver_note": null,
"task": "greeting",
"wall_ms": 20524
},
{
"baseline_failed": true,
"cost_usd": 0.2105145,
"energy_reliable": true,
"est_cloud_uj": 31648200000,
"input_tokens": 152031,
"microjoules": 0,
"note": "",
"output_tokens": 621,
"pass": 2,
"provenance": "ModelBased",
"resolved": true,
"solver": "claude",
"solver_note": null,
"task": "logic",
"wall_ms": 222810
},
{
"baseline_failed": true,
"cost_usd": 0.21362975,
"energy_reliable": true,
"est_cloud_uj": 31256600000,
"input_tokens": 148463,
"microjoules": 0,
"note": "",
"output_tokens": 782,
"pass": 2,
"provenance": "ModelBased",
"resolved": true,
"solver": "claude",
"solver_note": null,
"task": "rename",
"wall_ms": 19798
},
{
"baseline_failed": true,
"cost_usd": 0.208858,
"energy_reliable": true,
"est_cloud_uj": 30871400000,
"input_tokens": 147157,
"microjoules": 0,
"note": "",
"output_tokens": 720,
"pass": 2,
"provenance": "ModelBased",
"resolved": true,
"solver": "claude",
"solver_note": null,
"task": "typo",
"wall_ms": 22131
}
],
"suite": "tasks/mix"
}