575generated behavioral tests
98.4%official best score
93.6%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) | 20260519T215046Z |
Codex /goal | Paper prompt + Goal contract no internet | 93.6% | ok | 538/575 | n/a | 71 | 0.21h | not exported |
| 2 | GPT 5.5 (xhigh) | 20260517T094700Z |
Codex /goal | Mini-SWE-compatible no internet | 92.2% | ok | 530/575 | $5.01 | 83 | 0.27h | manifest · eval json · eval summary · usage audit |
| 3 | GPT 5.5 (xhigh) | 20260518T232006Z |
Codex /goal | Paper prompt + /goal no internet | 84.0% | ok | 483/575 | n/a | 68 | 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) · 20260519T215046Z
Public eval evidence has not been exported for this row yet.
GPT 5.5 (xhigh) · 20260517T094700Z
92.2% from 530/575 ProgramBench-scored tests. Raw public eval statuses: failure: 110, passed: 668.
Agent trace summary unavailable.
Likely miss class: behavioral mismatch, exact version/output mismatch.
c7d152868a35eval.tests.test_clog.test_from_conflicts_with_from_latest_tag (failure)c7d152868a35eval.tests.test_clog.test_changelog_conflicts_with_outfile (failure)c7d152868a35eval.tests.test_clog_errors.test_conflict_flags_error (failure)c7d152868a35eval.tests.test_clog_errors.test_from_and_from_latest_tag_conflict (failure)c7d152868a35eval.tests.test_clog_toml_heavy.test_toml_with_multiline_strings (failure)c7d152868a35eval.tests.test_simple_coverage_boost.test_conflicting_flags_error (failure)
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 | 98.4% | $11.10 | 97 |
| 2 | GPT 5.5 (high) | OpenAI | 95.5% | $2.73 | 41 |
| 3 | Claude Opus 4.7 (xhigh) | Anthropic | 93.2% | $18.38 | 333 |
| 4 | Claude Sonnet 4.6 | Anthropic | 93.0% | $18.61 | 453 |
| 5 | GPT 5.5 | OpenAI | 92.7% | $1.90 | 31 |
| 6 | Claude Opus 4.6 | Anthropic | 87.5% | $17.81 | 357 |
| 7 | Claude Opus 4.7 | Anthropic | 84.0% | $5.88 | 154 |
| 8 | Gemini 3.1 Pro | 76.3% | $2.23 | 182 | |
| 9 | GPT 5.4 | OpenAI | 65.0% | $0.32 | 12 |
| 10 | Claude Haiku 4.5 | Anthropic | 57.7% | $0.48 | 123 |
| 11 | GPT 5 mini | OpenAI | 32.5% | $0.02 | 12 |
| 12 | Gemini 3 Flash | 2.8% | $0.32 | 113 | |
| 13 | GPT 5.4 mini | OpenAI | 0.0% | $0.02 | 8 |