666generated behavioral tests
75.8%official best score
73.4%Codex best score
3Codex result rows
Codex Results by Model
| # | Model | Run | Agent | Mode | Score | Eval | Tests | Est. cost | Calls | Wall | Evidence |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | GPT 5.5 (xhigh) | 20260517T094700Z |
Codex /goal | Mini-SWE-compatible no internet | 73.4% | ok | 489/666 | $5.43 | 62 | 0.20h | manifest · eval json · eval summary · usage audit |
| 2 | GPT 5.5 (xhigh) | 20260519T215046Z |
Codex /goal | Paper prompt + Goal contract no internet | 73.4% | ok | 489/666 | n/a | 77 | 0.23h | not exported |
| 3 | GPT 5.5 (xhigh) | 20260518T232006Z |
Codex /goal | Paper prompt + /goal no internet | 65.3% | ok | 435/666 | n/a | 98 | 0.34h | not exported |
Why Scores Differ
Official ProgramBench rows are the public mini-SWE-agent submissions. GoalBench rows are separate Codex /goal submissions. Same model label does not mean the same agent, prompt, tool loop, or generated implementation. A GoalBench row is resolved only when the ProgramBench-scored pass rate is exactly 100%.
GPT 5.5 (xhigh) · 20260517T094700Z
73.4% from 489/666 ProgramBench-scored tests. Raw public eval statuses: error: 1, failure: 440, passed: 1547, skipped: 6.
Agent trace summary unavailable.
Likely miss class: behavioral mismatch.
17ed9c2b10e7eval.tests.test_args_coverage.test_invalid_priority (failure)17ed9c2b10e7eval.tests.test_comprehensive_cmdline.test_filter_hash_with_real_hash (failure)17ed9c2b10e7eval.tests.test_all_testdata.test_export_all_formats (failure)17ed9c2b10e7eval.tests.test_behavioral_strong.test_startday_shortcut (failure)17ed9c2b10e7eval.tests.test_behavioral_strong.test_query_appointments_specific_date (failure)17ed9c2b10e7eval.tests.test_additional_coverage.test_export_ical_explicit (failure)
GPT 5.5 (xhigh) · 20260519T215046Z
Public eval evidence has not been exported for this row yet.
GPT 5.5 (xhigh) · 20260518T232006Z
Public eval evidence has not been exported for this row yet.
Official ProgramBench Results by Model
| # | Model | Provider | Score | Cost | Calls |
|---|---|---|---|---|---|
| 1 | GPT 5.5 (xhigh) | OpenAI | 75.8% | $10.93 | 86 |
| 2 | GPT 5.5 (high) | OpenAI | 68.0% | $4.46 | 35 |
| 3 | GPT 5.5 | OpenAI | 54.8% | $1.85 | 27 |
| 4 | Claude Opus 4.6 | Anthropic | 53.8% | $12.74 | 248 |
| 5 | GPT 5.4 | OpenAI | 35.3% | $0.54 | 12 |
| 6 | Claude Haiku 4.5 | Anthropic | 35.0% | $0.83 | 127 |
| 7 | Claude Opus 4.7 (xhigh) | Anthropic | 33.9% | $3.30 | 84 |
| 8 | Gemini 3 Flash | 33.8% | $0.43 | 113 | |
| 9 | Claude Opus 4.7 | Anthropic | 33.5% | $2.88 | 80 |
| 10 | GPT 5.4 mini | OpenAI | 26.0% | $0.05 | 11 |
| 11 | Gemini 3.1 Pro | 23.1% | $1.69 | 84 | |
| 12 | GPT 5 mini | OpenAI | 17.0% | $0.01 | 8 |
| 13 | Claude Sonnet 4.6 | Anthropic | 7.1% | $31.57 | 528 |