425generated behavioral tests
82.4%official best score
84.2%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 | 84.2% | ok | 358/425 | n/a | 94 | 0.28h | not exported |
| 2 | GPT 5.5 (xhigh) | 20260517T094700Z |
Codex /goal | Mini-SWE-compatible no internet | 83.3% | ok | 354/425 | $5.64 | 107 | 0.22h | manifest · eval json · eval summary · usage audit |
| 3 | GPT 5.5 (xhigh) | 20260518T232006Z |
Codex /goal | Paper prompt + /goal no internet | 73.9% | ok | 314/425 | n/a | 47 | 0.20h | 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
83.3% from 354/425 ProgramBench-scored tests. Raw public eval statuses: failure: 95, passed: 461.
Agent trace summary unavailable.
Likely miss class: behavioral mismatch, exact version/output mismatch.
3487890d9158tests.test_basic_invocation.test_version_flag_equivalence (failure)3487890d9158tests.test_additional_coverage.test_package_with_build_error (failure)3487890d9158tests.test_display_options.test_failed_tests_always_shown (failure)3487890d9158tests.test_basic_invocation.test_version_flag_version (failure)3487890d9158tests.test_edge_cases.test_multiple_packages_mixed_results (failure)3487890d9158tests.test_failure_details.test_failure_with_stack_trace (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 (high) | OpenAI | 82.4% | $4.88 | 48 |
| 2 | GPT 5.5 (xhigh) | OpenAI | 81.6% | $7.81 | 81 |
| 3 | Claude Opus 4.7 (xhigh) | Anthropic | 80.2% | $15.52 | 168 |
| 4 | Claude Opus 4.6 | Anthropic | 77.6% | $13.53 | 179 |
| 5 | Claude Sonnet 4.6 | Anthropic | 76.0% | $21.05 | 361 |
| 6 | Claude Opus 4.7 | Anthropic | 72.0% | $2.33 | 69 |
| 7 | GPT 5.5 | OpenAI | 70.1% | $1.50 | 23 |
| 8 | Gemini 3 Flash | 56.9% | $0.26 | 69 | |
| 9 | Gemini 3.1 Pro | 51.8% | $1.05 | 55 | |
| 10 | Claude Haiku 4.5 | Anthropic | 51.5% | $0.62 | 84 |
| 11 | GPT 5.4 | OpenAI | 39.1% | $0.24 | 8 |
| 12 | GPT 5 mini | OpenAI | 28.7% | $0.03 | 13 |
| 13 | GPT 5.4 mini | OpenAI | 6.1% | $0.02 | 6 |