313generated behavioral tests
95.8%official best score
93.3%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 | 93.3% | ok | 292/313 | $5.75 | 92 | 0.34h | manifest · eval json · eval summary · usage audit |
| 2 | GPT 5.5 (xhigh) | 20260519T215046Z |
Codex /goal | Paper prompt + Goal contract no internet | 71.6% | ok | 224/313 | n/a | 89 | 0.31h | not exported |
| 3 | GPT 5.5 (xhigh) | 20260518T232006Z |
Codex /goal | Paper prompt + /goal no internet | 66.5% | ok | 208/313 | n/a | 76 | 0.33h | 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
93.3% from 292/313 ProgramBench-scored tests. Raw public eval statuses: failure: 24, passed: 345.
Agent trace summary unavailable.
Likely miss class: behavioral mismatch.
974a321c7e65eval.tests.test_encoding.test_encoding_case_insensitive_mixed (failure)974a321c7e65eval.tests.test_completion.test_completion_help (failure)974a321c7e65eval.tests.test_edge_cases.test_broken_pipe_handling (failure)974a321c7e65eval.tests.test_encoding.test_encoding_case_insensitive_upper (failure)25bb8e14ea1ftests.test_completion.test_completion_help_shows_usage (failure)25bb8e14ea1ftests.test_completion.test_completion_subcommand_with_extra_flag_after_shell (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 | 95.8% | $4.65 | 48 |
| 2 | Claude Opus 4.7 (xhigh) | Anthropic | 91.7% | $9.90 | 139 |
| 3 | Claude Opus 4.7 | Anthropic | 88.8% | $9.23 | 184 |
| 4 | Claude Opus 4.6 | Anthropic | 85.6% | $9.41 | 236 |
| 5 | GPT 5.5 (high) | OpenAI | 82.7% | $3.01 | 36 |
| 6 | GPT 5.5 | OpenAI | 75.7% | $0.88 | 13 |
| 7 | Gemini 3.1 Pro | 73.5% | $1.82 | 186 | |
| 8 | Claude Sonnet 4.6 | Anthropic | 71.6% | $14.48 | 348 |
| 9 | Gemini 3 Flash | 53.0% | $0.26 | 107 | |
| 10 | Claude Haiku 4.5 | Anthropic | 49.2% | $0.98 | 131 |
| 11 | GPT 5.4 | OpenAI | 38.0% | $0.25 | 11 |
| 12 | GPT 5 mini | OpenAI | 19.2% | $0.09 | 43 |
| 13 | GPT 5.4 mini | OpenAI | 0.0% | $0.06 | 11 |