465generated behavioral tests
80.9%official best score
67.3%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) | 20260519T215046Z |
Codex /goal | Paper prompt + Goal contract no internet | 67.3% | ok | 313/465 | n/a | 81 | 0.31h | not exported |
| 2 | GPT 5.5 (xhigh) | 20260517T094700Z |
Codex /goal | Mini-SWE-compatible no internet | 58.3% | ok | 271/465 | $3.49 | 61 | 0.20h | manifest · eval json · eval summary · usage audit |
| 3 | GPT 5.5 (xhigh) | 20260518T232006Z |
Codex /goal | Paper prompt + /goal no internet | 57.6% | ok | 268/465 | n/a | 64 | 0.23h | 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
58.3% from 271/465 ProgramBench-scored tests. Raw public eval statuses: failure: 196, passed: 291, skipped: 1.
Agent trace summary unavailable.
Likely miss class: behavioral mismatch, exact version/output mismatch.
da1ea1b44b41tests.test_braille.test_braille_threshold_256_out_of_range_error (failure)da1ea1b44b41tests.test_basic_conversion.test_checkerboard_pattern_preservation (failure)da1ea1b44b41tests.test_basic_conversion.test_stdin_piped_input (failure)da1ea1b44b41tests.test_basic_conversion.test_dimensions_wrong_count_rejected (failure)da1ea1b44b41tests.test_basic_conversion.test_red_color_brightness_mapping (failure)da1ea1b44b41tests.test_braille.test_braille_with_grayscale_flag (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 | 80.9% | $10.29 | 95 |
| 2 | GPT 5.5 (high) | OpenAI | 77.2% | $4.32 | 48 |
| 3 | Claude Opus 4.7 (xhigh) | Anthropic | 76.6% | $27.07 | 238 |
| 4 | Claude Opus 4.6 | Anthropic | 64.1% | $15.75 | 278 |
| 5 | Claude Opus 4.7 | Anthropic | 58.9% | $7.97 | 178 |
| 6 | GPT 5.5 | OpenAI | 56.3% | $1.11 | 16 |
| 7 | Claude Sonnet 4.6 | Anthropic | 53.8% | $36.37 | 584 |
| 8 | Gemini 3 Flash | 38.7% | $0.40 | 92 | |
| 9 | GPT 5.4 | OpenAI | 28.0% | $0.28 | 7 |
| 10 | Claude Haiku 4.5 | Anthropic | 16.6% | $0.81 | 135 |
| 11 | GPT 5 mini | OpenAI | 12.5% | $0.03 | 16 |
| 12 | Gemini 3.1 Pro | 0.0% | $1.30 | 76 | |
| 13 | GPT 5.4 mini | OpenAI | 0.0% | $0.02 | 8 |