722generated behavioral tests
96.0%official best score
92.4%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 | 92.4% | ok | 667/722 | $4.29 | 70 | 0.22h | manifest · eval json · eval summary · usage audit |
| 2 | GPT 5.5 (xhigh) | 20260518T232006Z |
Codex /goal | Paper prompt + /goal no internet | 86.1% | ok | 622/722 | n/a | 84 | 0.20h | not exported |
| 3 | GPT 5.5 (xhigh) | 20260519T215046Z |
Codex /goal | Paper prompt + Goal contract no internet | 75.1% | ok | 542/722 | n/a | 49 | 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) · 20260517T094700Z
92.4% from 667/722 ProgramBench-scored tests. Raw public eval statuses: failure: 95, passed: 832, skipped: 2.
Agent trace summary unavailable.
Likely miss class: behavioral mismatch.
124d12a3e000tests.test_edge_cases.test_project_filter_with_question_mark (failure)124d12a3e000tests.test_edge_cases.test_project_filter_with_wildcard (failure)124d12a3e000tests.test_last_projects_search.test_search_wildcard_asterisk (failure)124d12a3e000tests.test_last_projects_search.test_search_wildcard_question (failure)116a722f8c6etests.test_edit_and_edge_cases.test_wildcard_in_project_filter (failure)116a722f8c6etests.test_format_and_display.test_wildcard_project_filter_question_mark (failure)
GPT 5.5 (xhigh) · 20260518T232006Z
Public eval evidence has not been exported for this row yet.
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 | 96.0% | $9.37 | 89 |
| 2 | GPT 5.5 (high) | OpenAI | 90.4% | $4.12 | 39 |
| 3 | Claude Opus 4.7 | Anthropic | 87.3% | $13.55 | 219 |
| 4 | Claude Opus 4.7 (xhigh) | Anthropic | 84.6% | $17.84 | 250 |
| 5 | Claude Sonnet 4.6 | Anthropic | 84.3% | $13.63 | 283 |
| 6 | Claude Opus 4.6 | Anthropic | 80.9% | $10.92 | 274 |
| 7 | GPT 5.5 | OpenAI | 64.5% | $0.94 | 14 |
| 8 | Gemini 3.1 Pro | 62.5% | $1.58 | 99 | |
| 9 | GPT 5.4 | OpenAI | 58.6% | $0.42 | 12 |
| 10 | Gemini 3 Flash | 51.1% | $0.25 | 86 | |
| 11 | Claude Haiku 4.5 | Anthropic | 48.3% | $0.89 | 149 |
| 12 | GPT 5.4 mini | OpenAI | 23.3% | $0.06 | 10 |
| 13 | GPT 5 mini | OpenAI | 18.7% | $0.04 | 19 |