683generated behavioral tests
94.0%official best score
90.8%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) | 20260518T232006Z |
Codex /goal | Paper prompt + /goal no internet | 90.8% | ok | 620/683 | n/a | 85 | 0.19h | not exported |
| 2 | GPT 5.5 (xhigh) | 20260517T094700Z |
Codex /goal | Mini-SWE-compatible no internet | 89.9% | ok | 614/683 | $5.91 | 112 | 0.27h | manifest · eval json · eval summary · usage audit |
| 3 | GPT 5.5 (xhigh) | 20260519T215046Z |
Codex /goal | Paper prompt + Goal contract no internet | 86.4% | ok | 590/683 | n/a | 62 | 0.19h | 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) · 20260518T232006Z
Public eval evidence has not been exported for this row yet.
GPT 5.5 (xhigh) · 20260517T094700Z
89.9% from 614/683 ProgramBench-scored tests. Raw public eval statuses: failure: 70, passed: 670.
Agent trace summary unavailable.
Likely miss class: behavioral mismatch, exact version/output mismatch.
8210eaf4f82etests.test_edge_cases.test_whitespace_in_replacement (failure)8210eaf4f82etests.test_edge_cases.test_replace_with_period (failure)8210eaf4f82etests.test_edge_cases.test_empty_replacement (failure)7877cb75132etests.test_errors.test_invalid_regex_unclosed_bracket (failure)7877cb75132etests.test_advanced_regex.test_backreference_in_pattern (failure)7877cb75132etests.test_advanced_regex.test_replace_limit_unlimited (failure)
GPT 5.5 (xhigh) · 20260519T215046Z
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 | 94.0% | $6.29 | 73 |
| 2 | GPT 5.5 (high) | OpenAI | 92.7% | $2.75 | 50 |
| 3 | Claude Opus 4.7 (xhigh) | Anthropic | 89.6% | $50.30 | 594 |
| 4 | Claude Opus 4.7 | Anthropic | 82.1% | $4.44 | 154 |
| 5 | GPT 5.5 | OpenAI | 81.7% | $1.10 | 19 |
| 6 | Gemini 3.1 Pro | 79.8% | $1.95 | 184 | |
| 7 | Claude Sonnet 4.6 | Anthropic | 78.0% | $20.12 | 374 |
| 8 | GPT 5.4 | OpenAI | 67.9% | $0.37 | 14 |
| 9 | Claude Opus 4.6 | Anthropic | 66.6% | $11.86 | 257 |
| 10 | GPT 5.4 mini | OpenAI | 25.9% | $0.04 | 18 |
| 11 | Claude Haiku 4.5 | Anthropic | 21.4% | $1.33 | 154 |
| 12 | Gemini 3 Flash | 3.8% | $0.26 | 107 | |
| 13 | GPT 5 mini | OpenAI | 3.8% | $0.03 | 24 |