339generated behavioral tests
88.2%official best score
71.1%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 | 71.1% | ok | 241/339 | n/a | 91 | 0.23h | not exported |
| 2 | GPT 5.5 (xhigh) | 20260517T094700Z |
Codex /goal | Mini-SWE-compatible no internet | 69.9% | ok | 237/339 | $1.84 | 35 | 0.12h | manifest · eval json · eval summary · usage audit |
| 3 | GPT 5.5 (xhigh) | 20260518T232006Z |
Codex /goal | Paper prompt + /goal no internet | 68.4% | ok | 232/339 | n/a | 56 | 0.14h | 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) · 20260519T215046Z
Public eval evidence has not been exported for this row yet.
GPT 5.5 (xhigh) · 20260517T094700Z
69.9% from 237/339 ProgramBench-scored tests. Raw public eval statuses: failure: 121, passed: 530, skipped: 4.
Agent trace summary unavailable.
Likely miss class: behavioral mismatch, exact version/output mismatch, terminal rendering mismatch.
aa573607d388eval.tests.test_tui_execution.test_tui_with_watch_interval (failure)aa573607d388eval.tests.test_tui_execution.test_tui_with_buffer_option (failure)aa573607d388eval.tests.test_tui_execution.test_tui_runs_with_localhost (failure)aa573607d388eval.tests.test_tui_execution.test_tui_with_color (failure)aa573607d388eval.tests.test_tui_execution.test_tui_ipv4_mode (failure)aa573607d388eval.tests.test_tui_execution.test_tui_with_simple_graphics (failure)
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 | 88.2% | $8.13 | 76 |
| 2 | GPT 5.5 (high) | OpenAI | 85.0% | $3.77 | 49 |
| 3 | Claude Opus 4.6 | Anthropic | 78.5% | $10.87 | 211 |
| 4 | Claude Opus 4.7 (xhigh) | Anthropic | 77.3% | $5.92 | 120 |
| 5 | Claude Opus 4.7 | Anthropic | 72.3% | $1.89 | 70 |
| 6 | GPT 5.5 | OpenAI | 68.7% | $0.82 | 16 |
| 7 | Claude Sonnet 4.6 | Anthropic | 58.7% | $11.38 | 271 |
| 8 | GPT 5.4 | OpenAI | 49.6% | $0.18 | 8 |
| 9 | Gemini 3.1 Pro | 36.9% | $0.66 | 33 | |
| 10 | Gemini 3 Flash | 29.8% | $0.22 | 64 | |
| 11 | GPT 5 mini | OpenAI | 21.8% | $0.01 | 7 |
| 12 | GPT 5.4 mini | OpenAI | 20.9% | $0.03 | 18 |
| 13 | Claude Haiku 4.5 | Anthropic | 16.2% | $0.48 | 113 |