341generated behavioral tests
83.3%official best score
58.9%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 | 58.9% | ok | 201/341 | $4.90 | 69 | 0.28h | manifest · eval json · eval summary · usage audit |
| 2 | GPT 5.5 (xhigh) | 20260519T215046Z |
Codex /goal | Paper prompt + Goal contract no internet | 49.0% | ok | 167/341 | n/a | 63 | 0.22h | not exported |
| 3 | GPT 5.5 (xhigh) | 20260518T232006Z |
Codex /goal | Paper prompt + /goal no internet | 34.6% | ok | 118/341 | n/a | 38 | 0.21h | 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
58.9% from 201/341 ProgramBench-scored tests. Raw public eval statuses: failure: 203, passed: 325, skipped: 4.
Agent trace summary unavailable.
Likely miss class: behavioral mismatch.
3b27b965f1d5eval.tests.test_errcheck.test_output_contains_code_snippet (failure)3b27b965f1d5eval.tests.test_errcheck.test_exclude_with_comments (failure)3b27b965f1d5eval.tests.test_errcheck.test_mod_flag (failure)3b27b965f1d5eval.tests.test_errcheck.test_testdata_custom_error_types (failure)3b27b965f1d5eval.tests.test_errcheck.test_detect_unchecked_errors_in_testdata (failure)3b27b965f1d5eval.tests.test_errcheck.test_verbose_output_includes_function_context (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 | 83.3% | $8.89 | 113 |
| 2 | Claude Opus 4.6 | Anthropic | 80.4% | $10.88 | 225 |
| 3 | GPT 5.5 (high) | OpenAI | 80.1% | $3.59 | 53 |
| 4 | Claude Opus 4.7 (xhigh) | Anthropic | 66.0% | $29.50 | 395 |
| 5 | Claude Opus 4.7 | Anthropic | 56.9% | $6.93 | 187 |
| 6 | Claude Sonnet 4.6 | Anthropic | 56.6% | $17.09 | 426 |
| 7 | Gemini 3.1 Pro | 50.4% | $2.07 | 135 | |
| 8 | GPT 5.4 | OpenAI | 48.7% | $0.34 | 14 |
| 9 | GPT 5.5 | OpenAI | 47.8% | $1.55 | 22 |
| 10 | Gemini 3 Flash | 36.7% | $0.22 | 112 | |
| 11 | GPT 5.4 mini | OpenAI | 22.6% | $0.03 | 14 |
| 12 | Claude Haiku 4.5 | Anthropic | 21.1% | $1.49 | 216 |
| 13 | GPT 5 mini | OpenAI | 10.9% | $0.01 | 8 |