817generated behavioral tests
73.2%official best score
74.7%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 | 74.7% | ok | 610/817 | $5.40 | 81 | 0.22h | manifest · eval json · eval summary · usage audit |
| 2 | GPT 5.5 (xhigh) | 20260519T215046Z |
Codex /goal | Paper prompt + Goal contract no internet | 67.6% | ok | 552/817 | n/a | 83 | 0.26h | not exported |
| 3 | GPT 5.5 (xhigh) | 20260518T232006Z |
Codex /goal | Paper prompt + /goal no internet | 55.0% | ok | 449/817 | n/a | 43 | 0.15h | 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
74.7% from 610/817 ProgramBench-scored tests. Raw public eval statuses: failure: 209, passed: 879.
Agent trace summary unavailable.
Likely miss class: behavioral mismatch, interactive terminal behavior mismatch.
6393fbdb1dffeval.tests.test_date_time_filters.test_date_iso_format (failure)6393fbdb1dffeval.tests.test_display_options.test_color_yes (failure)c0926ca6510eeval.tests.test_rhit.test_color_yes (failure)c0926ca6510eeval.tests.test_rhit_coverage_boost.test_date_parsing_various_formats (failure)c0926ca6510eeval.tests.test_rhit_coverage_boost.test_printer_color_variations (failure)5acb4aa8bf92tests.test_basic.test_help_message (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 | 73.2% | $12.36 | 78 |
| 2 | GPT 5.5 (high) | OpenAI | 70.1% | $4.62 | 43 |
| 3 | Claude Opus 4.7 (xhigh) | Anthropic | 57.5% | $1.65 | 34 |
| 4 | Claude Opus 4.7 | Anthropic | 53.2% | $1.18 | 34 |
| 5 | Claude Sonnet 4.6 | Anthropic | 52.5% | $40.64 | 458 |
| 6 | Claude Opus 4.6 | Anthropic | 52.4% | $15.70 | 230 |
| 7 | GPT 5.5 | OpenAI | 44.8% | $1.74 | 17 |
| 8 | GPT 5.4 | OpenAI | 43.2% | $0.23 | 7 |
| 9 | Gemini 3 Flash | 38.4% | $0.40 | 68 | |
| 10 | Gemini 3.1 Pro | 31.1% | $3.17 | 135 | |
| 11 | Claude Haiku 4.5 | Anthropic | 28.5% | $1.08 | 134 |
| 12 | GPT 5 mini | OpenAI | 7.8% | $0.02 | 15 |
| 13 | GPT 5.4 mini | OpenAI | 1.3% | $0.07 | 17 |