480generated behavioral tests
81.9%official best score
42.1%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 | 42.1% | ok | 202/480 | $4.20 | 85 | 0.22h | manifest · eval json · eval summary · usage audit |
| 2 | GPT 5.5 (xhigh) | 20260519T215046Z |
Codex /goal | Paper prompt + Goal contract no internet | 35.8% | ok | 172/480 | n/a | 91 | 0.24h | not exported |
| 3 | GPT 5.5 (xhigh) | 20260518T232006Z |
Codex /goal | Paper prompt + /goal no internet | 33.8% | ok | 162/480 | n/a | 49 | 0.18h | 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
42.1% from 202/480 ProgramBench-scored tests. Raw public eval statuses: failure: 324, passed: 340, skipped: 5.
Agent trace summary unavailable.
Likely miss class: behavioral mismatch, exact version/output mismatch.
c784b3264408eval.tests.test_wrapcheck.test_config_ignore_package_globs (failure)c784b3264408eval.tests.test_wrapcheck.test_config_invalid_yaml (failure)c784b3264408eval.tests.test_wrapcheck.test_json_output_format (failure)c784b3264408eval.tests.test_wrapcheck.test_analyze_package_with_issues (failure)c784b3264408eval.tests.test_wrapcheck.test_detect_unwrapped_json_error (failure)c784b3264408eval.tests.test_wrapcheck.test_config_ignore_sig_regexps (failure)
GPT 5.5 (xhigh) · 20260519T215046Z
Public eval evidence has not been exported for this row yet.
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 (high) | OpenAI | 81.9% | $2.12 | 27 |
| 2 | Claude Opus 4.6 | Anthropic | 80.8% | $10.87 | 250 |
| 3 | GPT 5.5 (xhigh) | OpenAI | 65.0% | $6.65 | 85 |
| 4 | GPT 5.5 | OpenAI | 58.8% | $1.07 | 19 |
| 5 | Gemini 3.1 Pro | 40.0% | $0.96 | 83 | |
| 6 | Claude Sonnet 4.6 | Anthropic | 36.0% | $11.82 | 325 |
| 7 | Claude Opus 4.7 (xhigh) | Anthropic | 25.8% | $10.56 | 164 |
| 8 | Claude Opus 4.7 | Anthropic | 24.6% | $6.25 | 148 |
| 9 | Gemini 3 Flash | 23.3% | $0.30 | 114 | |
| 10 | Claude Haiku 4.5 | Anthropic | 20.6% | $1.12 | 153 |
| 11 | GPT 5.4 | OpenAI | 9.2% | $0.31 | 13 |
| 12 | GPT 5 mini | OpenAI | 8.1% | $0.03 | 22 |
| 13 | GPT 5.4 mini | OpenAI | 1.0% | $0.03 | 21 |