380generated behavioral tests
66.8%official best score
64.7%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 | 64.7% | ok | 246/380 | n/a | 55 | 0.22h | not exported |
| 2 | GPT 5.5 (xhigh) | 20260517T094700Z |
Codex /goal | Mini-SWE-compatible no internet | 64.5% | ok | 245/380 | $5.93 | 82 | 0.30h | manifest · eval json · eval summary · usage audit |
| 3 | GPT 5.5 (xhigh) | 20260519T215046Z |
Codex /goal | Paper prompt + Goal contract no internet | 62.1% | ok | 236/380 | n/a | 61 | 0.17h | 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
64.5% from 245/380 ProgramBench-scored tests. Raw public eval statuses: error: 1, failure: 134, passed: 313, skipped: 6.
Agent trace summary unavailable.
Likely miss class: behavioral mismatch, exact version/output mismatch.
4e26fc9205c7eval.tests.test_error_handling.test_corrupted_obj_file (failure)4e26fc9205c7eval.tests.test_error_handling.test_invalid_file_extension_error (failure)4e26fc9205c7eval.tests.test_input_files.test_unsupported_extension (failure)4e26fc9205c7eval.tests.test_rotation_flags.test_pitch_rotation_short (failure)4e26fc9205c7eval.tests.test_rotation_flags.test_negative_rotation_long_flag (failure)718b6511c1b3eval.tests.test_input_handling.test_invalid_extension (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 | 66.8% | $7.18 | 57 |
| 2 | GPT 5.5 (high) | OpenAI | 65.3% | $3.62 | 38 |
| 3 | GPT 5.5 | OpenAI | 57.9% | $1.46 | 20 |
| 4 | Claude Opus 4.7 (xhigh) | Anthropic | 56.3% | $16.22 | 187 |
| 5 | Claude Opus 4.6 | Anthropic | 52.6% | $16.83 | 237 |
| 6 | Claude Opus 4.7 | Anthropic | 47.4% | $2.46 | 58 |
| 7 | Claude Sonnet 4.6 | Anthropic | 45.5% | $83.70 | 578 |
| 8 | GPT 5.4 | OpenAI | 36.6% | $0.41 | 10 |
| 9 | Gemini 3 Flash | 28.9% | $0.46 | 95 | |
| 10 | GPT 5.4 mini | OpenAI | 25.3% | $0.05 | 10 |
| 11 | Claude Haiku 4.5 | Anthropic | 24.2% | $0.95 | 126 |
| 12 | GPT 5 mini | OpenAI | 23.2% | $0.03 | 20 |
| 13 | Gemini 3.1 Pro | 0.0% | $4.94 | 151 |