882generated behavioral tests
23.6%official best score
24.6%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) | 20260517T094700Z |
Codex /goal | Mini-SWE-compatible no internet | 24.6% | ok | 217/882 | $4.71 | 77 | 0.23h | manifest · eval json · eval summary · usage audit |
| 2 | GPT 5.5 (xhigh) | 20260519T215046Z |
Codex /goal | Paper prompt + Goal contract no internet | 21.2% | ok | 187/882 | n/a | 57 | 0.22h | not exported |
| 3 | GPT 5.5 (xhigh) | 20260518T232006Z |
Codex /goal | Paper prompt + /goal no internet | 20.5% | ok | 181/882 | n/a | 55 | 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
24.6% from 217/882 ProgramBench-scored tests. Raw public eval statuses: failure: 670, passed: 225.
Agent trace summary unavailable.
Likely miss class: behavioral mismatch, interactive terminal behavior mismatch.
5946c18eda4ceval.tests.test_debug_query.test_debug_query_ast_without_color (failure)5946c18eda4ceval.tests.test_debug_query.test_debug_query_ast_go (failure)5946c18eda4ceval.tests.test_config_validation.test_undefined_metavar_in_transform_section (failure)5946c18eda4ceval.tests.test_errors.test_missing_required_rewrite (failure)5946c18eda4ceval.tests.test_error_diagnostics.test_parse_test_error (failure)5946c18eda4ceval.tests.test_completions.test_bash_completions_exact_golden (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 (xhigh) | OpenAI | 23.6% | $8.63 | 67 |
| 2 | GPT 5.5 (high) | OpenAI | 20.2% | $3.93 | 35 |
| 3 | GPT 5.5 | OpenAI | 12.7% | $1.48 | 21 |
| 4 | Claude Opus 4.6 | Anthropic | 11.9% | $11.94 | 322 |
| 5 | Claude Opus 4.7 (xhigh) | Anthropic | 9.5% | $5.83 | 123 |
| 6 | GPT 5.4 | OpenAI | 9.4% | $0.41 | 11 |
| 7 | Claude Sonnet 4.6 | Anthropic | 8.0% | $28.64 | 621 |
| 8 | Gemini 3.1 Pro | 5.9% | $0.79 | 66 | |
| 9 | Claude Opus 4.7 | Anthropic | 3.9% | $1.45 | 46 |
| 10 | Gemini 3 Flash | 2.6% | $0.27 | 70 | |
| 11 | Claude Haiku 4.5 | Anthropic | 2.6% | $0.49 | 92 |
| 12 | GPT 5 mini | OpenAI | 1.4% | $0.04 | 16 |
| 13 | GPT 5.4 mini | OpenAI | 0.0% | $0.05 | 38 |