564generated behavioral tests
98.2%official best score
75.9%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 | 75.9% | ok | 428/564 | $4.24 | 54 | 0.19h | manifest · eval json · eval summary · usage audit |
| 2 | GPT 5.5 (xhigh) | 20260518T232006Z |
Codex /goal | Paper prompt + /goal no internet | 71.5% | ok | 403/564 | n/a | 57 | 0.20h | not exported |
| 3 | GPT 5.5 (xhigh) | 20260519T215046Z |
Codex /goal | Paper prompt + Goal contract no internet | 41.3% | ok | 233/564 | n/a | 46 | 0.16h | 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
75.9% from 428/564 ProgramBench-scored tests. Raw public eval statuses: failure: 142, passed: 503, skipped: 1.
Agent trace summary unavailable.
Likely miss class: behavioral mismatch, exact version/output mismatch.
f75c6d083a5ceval.tests.test_output_generation.test_html_contains_css (failure)a2367572e7fctests.test_elf_variants.test_elf_with_abi_version (failure)0b64bd84aa2btests.test_edge_cases.test_section_names_with_html_special_characters (failure)0b64bd84aa2btests.test_cli_errors.test_concurrent_different_files (failure)0b64bd84aa2btests.test_cli_errors.test_special_chars_in_filename (failure)0b64bd84aa2btests.test_cli_errors.test_output_filename_pattern (failure)
GPT 5.5 (xhigh) · 20260518T232006Z
Public eval evidence has not been exported for this row yet.
GPT 5.5 (xhigh) · 20260519T215046Z
Public eval evidence has not been exported for this row yet.
Official ProgramBench Results by Model
| # | Model | Provider | Score | Cost | Calls |
|---|---|---|---|---|---|
| 1 | Claude Opus 4.7 | Anthropic | 98.2% | $23.52 | 271 |
| 2 | Claude Opus 4.6 | Anthropic | 92.9% | $15.79 | 251 |
| 3 | GPT 5.5 (xhigh) | OpenAI | 87.6% | $7.57 | 73 |
| 4 | Claude Opus 4.7 (xhigh) | Anthropic | 85.1% | $19.73 | 243 |
| 5 | GPT 5.5 (high) | OpenAI | 82.8% | $9.37 | 65 |
| 6 | Claude Sonnet 4.6 | Anthropic | 81.6% | $23.39 | 403 |
| 7 | GPT 5.5 | OpenAI | 65.4% | $0.96 | 15 |
| 8 | Gemini 3 Flash | 53.2% | $0.51 | 139 | |
| 9 | Gemini 3.1 Pro | 45.2% | $2.90 | 140 | |
| 10 | Claude Haiku 4.5 | Anthropic | 27.5% | $0.46 | 89 |
| 11 | GPT 5.4 mini | OpenAI | 19.1% | $0.03 | 9 |
| 12 | GPT 5 mini | OpenAI | 2.7% | $0.04 | 26 |
| 13 | GPT 5.4 | OpenAI | 2.5% | $0.19 | 10 |