1108generated behavioral tests
43.1%official best score
34.0%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) | 20260519T215046Z |
Codex /goal | Paper prompt + Goal contract no internet | 34.0% | ok | 377/1108 | n/a | 56 | 0.21h | not exported |
| 2 | GPT 5.5 (xhigh) | 20260518T232006Z |
Codex /goal | Paper prompt + /goal no internet | 26.8% | ok | 297/1108 | n/a | 44 | 0.14h | not exported |
| 3 | GPT 5.5 (xhigh) | 20260517T094700Z |
Codex /goal | Mini-SWE-compatible no internet | 10.6% | ok | 117/1108 | $5.66 | 99 | 0.26h | manifest · eval json · eval summary · usage audit |
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) · 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.
GPT 5.5 (xhigh) · 20260517T094700Z
10.6% from 117/1108 ProgramBench-scored tests. Raw public eval statuses: error: 74, failure: 78, not_run: 1185, passed: 134.
Agent trace summary unavailable.
Likely miss class: behavioral mismatch, interactive terminal behavior mismatch.
a94a7d27d002eval.tests.test_advanced_features.test_list_format_full (error)a94a7d27d002eval.tests.test_advanced_features.test_new_with_group (error)a94a7d27d002eval.tests.test_configuration.test_list_multiple_match_terms (error)a94a7d27d002eval.tests.test_advanced_features.test_tag_list_format_full (error)a94a7d27d002eval.tests.test_advanced_features.test_list_recursive_links (error)a94a7d27d002eval.tests.test_advanced_features.test_list_mentioned_by_filter (error)
Official ProgramBench Results by Model
| # | Model | Provider | Score | Cost | Calls |
|---|---|---|---|---|---|
| 1 | Claude Sonnet 4.6 | Anthropic | 43.1% | $24.25 | 494 |
| 2 | GPT 5.4 | OpenAI | 17.0% | $0.30 | 8 |
| 3 | GPT 5.5 (high) | OpenAI | 16.5% | $4.05 | 51 |
| 4 | GPT 5.5 (xhigh) | OpenAI | 16.2% | $7.05 | 80 |
| 5 | Claude Opus 4.7 (xhigh) | Anthropic | 16.1% | $9.09 | 113 |
| 6 | GPT 5.5 | OpenAI | 15.3% | $1.30 | 22 |
| 7 | Claude Opus 4.7 | Anthropic | 14.0% | $1.52 | 59 |
| 8 | Gemini 3 Flash | 13.9% | $0.21 | 62 | |
| 9 | Gemini 3.1 Pro | 12.1% | $1.75 | 139 | |
| 10 | GPT 5 mini | OpenAI | 7.5% | $0.02 | 12 |
| 11 | GPT 5.4 mini | OpenAI | 5.1% | $0.03 | 9 |
| 12 | Claude Opus 4.6 | Anthropic | 1.2% | $14.07 | 286 |
| 13 | Claude Haiku 4.5 | Anthropic | 1.2% | $0.59 | 91 |