792generated behavioral tests
86.5%official best score
83.2%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) | 20260517T094700Z |
Codex /goal | Mini-SWE-compatible no internet | 83.2% | ok | 659/792 | $5.63 | 99 | 0.26h | manifest · eval json · eval summary · usage audit |
| 2 | GPT 5.5 (xhigh) | 20260519T215046Z |
Codex /goal | Paper prompt + Goal contract no internet | 44.9% | ok | 356/792 | n/a | 50 | 0.20h | not exported |
| 3 | GPT 5.5 (xhigh) | 20260518T232006Z |
Codex /goal | Paper prompt + /goal no internet | 44.6% | ok | 353/792 | n/a | 42 | 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
83.2% from 659/792 ProgramBench-scored tests. Raw public eval statuses: failure: 136, passed: 741, skipped: 2.
Agent trace summary unavailable.
Likely miss class: behavioral mismatch.
4bb8d103b799eval.tests.test_css_selector.test_css_attr_requires_query (failure)4bb8d103b799eval.tests.test_edge_cases.test_malformed_xml (failure)4bb8d103b799eval.tests.test_config.test_config_tab_setting (failure)4bb8d103b799eval.tests.test_json_output.test_json_to_json_pass_through (failure)4bb8d103b799eval.tests.test_xml_formatting.test_format_xml_zero_indent (failure)cd8785aceabaeval.tests.test_additional_coverage.test_json_to_json_with_formatting (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 | 86.5% | $10.18 | 99 |
| 2 | Claude Opus 4.7 (xhigh) | Anthropic | 79.0% | $12.57 | 214 |
| 3 | Claude Opus 4.6 | Anthropic | 75.9% | $13.82 | 356 |
| 4 | GPT 5.5 (high) | OpenAI | 75.8% | $3.52 | 37 |
| 5 | Claude Sonnet 4.6 | Anthropic | 75.0% | $26.84 | 625 |
| 6 | GPT 5.5 | OpenAI | 72.7% | $0.87 | 16 |
| 7 | Gemini 3 Flash | 63.5% | $0.31 | 111 | |
| 8 | Gemini 3.1 Pro | 63.1% | $2.94 | 158 | |
| 9 | GPT 5.4 | OpenAI | 57.8% | $0.66 | 11 |
| 10 | Claude Haiku 4.5 | Anthropic | 37.8% | $0.98 | 103 |
| 11 | GPT 5.4 mini | OpenAI | 37.0% | $0.06 | 13 |
| 12 | Claude Opus 4.7 | Anthropic | 2.5% | $11.86 | 227 |
| 13 | GPT 5 mini | OpenAI | 1.5% | $0.03 | 15 |