1196generated behavioral tests
96.1%official best score
93.3%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 | 93.3% | ok | 1116/1196 | n/a | 44 | 0.19h | not exported |
| 2 | GPT 5.5 (xhigh) | 20260517T094700Z |
Codex /goal | Mini-SWE-compatible no internet | 92.6% | ok | 1108/1196 | $5.68 | 86 | 0.26h | manifest · eval json · eval summary · usage audit |
| 3 | GPT 5.5 (xhigh) | 20260519T215046Z |
Codex /goal | Paper prompt + Goal contract no internet | 86.5% | ok | 1035/1196 | n/a | 40 | 0.16h | 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
92.6% from 1108/1196 ProgramBench-scored tests. Raw public eval statuses: failure: 88, passed: 1157, skipped: 4.
Agent trace summary unavailable.
Likely miss class: behavioral mismatch.
e73d7b48bf09eval.tests.test_error_combinations.test_json_with_no_join_conflict (failure)e73d7b48bf09eval.tests.test_error_combinations.test_json_with_replace_conflict (failure)e73d7b48bf09eval.tests.test_error_combinations.test_characters_with_no_join_conflict (failure)e73d7b48bf09eval.tests.test_error_combinations.test_fixed_memory_zero_error (failure)e57edf5017f8eval.tests.test_field_selection.test_format_with_newline (failure)e57edf5017f8eval.tests.test_field_selection.test_format_with_tab (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 | 96.1% | $12.77 | 105 |
| 2 | Claude Opus 4.7 | Anthropic | 92.7% | $8.58 | 165 |
| 3 | Claude Opus 4.6 | Anthropic | 90.5% | $16.11 | 317 |
| 4 | GPT 5.5 (high) | OpenAI | 89.5% | $4.67 | 47 |
| 5 | Claude Opus 4.7 (xhigh) | Anthropic | 88.6% | $15.11 | 188 |
| 6 | Claude Sonnet 4.6 | Anthropic | 88.1% | $16.19 | 506 |
| 7 | GPT 5.5 | OpenAI | 60.7% | $0.89 | 13 |
| 8 | Claude Haiku 4.5 | Anthropic | 58.9% | $0.94 | 182 |
| 9 | GPT 5.4 | OpenAI | 53.8% | $0.30 | 10 |
| 10 | GPT 5 mini | OpenAI | 37.0% | $0.02 | 14 |
| 11 | Gemini 3.1 Pro | 17.1% | $0.94 | 87 | |
| 12 | Gemini 3 Flash | 1.5% | $0.28 | 82 | |
| 13 | GPT 5.4 mini | OpenAI | 0.0% | $0.04 | 22 |