385generated behavioral tests
86.2%official best score
75.3%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 | 75.3% | ok | 290/385 | $2.51 | 55 | 0.15h | manifest · eval json · eval summary · usage audit |
| 2 | GPT 5.5 (xhigh) | 20260519T215046Z |
Codex /goal | Paper prompt + Goal contract no internet | 74.0% | ok | 285/385 | n/a | 47 | 0.12h | not exported |
| 3 | GPT 5.5 (xhigh) | 20260518T232006Z |
Codex /goal | Paper prompt + /goal no internet | 72.7% | ok | 280/385 | n/a | 32 | 0.12h | 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
75.3% from 290/385 ProgramBench-scored tests. Raw public eval statuses: failure: 126, passed: 577, skipped: 1.
Agent trace summary unavailable.
Likely miss class: behavioral mismatch, interactive terminal behavior mismatch, terminal rendering mismatch.
df37316a82f6tests.test_cli.test_invalid_glob_pattern_error (failure)df37316a82f6tests.test_editor.test_env_var_editor_with_path (failure)df37316a82f6tests.test_editor.test_custom_command_overrides_igrep_custom_editor (failure)df37316a82f6tests.test_keymap_popup.test_keymap_popup_closes_with_f1_key (failure)df37316a82f6tests.test_keymap_popup.test_keymap_popup_scroll_left (failure)df37316a82f6tests.test_keymap_popup.test_keymap_popup_scroll_to_bottom_then_top (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 | 86.2% | $4.92 | 51 |
| 2 | GPT 5.5 (high) | OpenAI | 75.8% | $2.19 | 32 |
| 3 | Claude Opus 4.7 (xhigh) | Anthropic | 75.6% | $8.64 | 125 |
| 4 | GPT 5.5 | OpenAI | 74.0% | $1.02 | 20 |
| 5 | Claude Opus 4.7 | Anthropic | 73.5% | $1.38 | 54 |
| 6 | Gemini 3.1 Pro | 61.8% | $0.74 | 53 | |
| 7 | Claude Sonnet 4.6 | Anthropic | 60.5% | $10.26 | 279 |
| 8 | Gemini 3 Flash | 56.9% | $0.21 | 49 | |
| 9 | Claude Haiku 4.5 | Anthropic | 53.5% | $0.43 | 76 |
| 10 | GPT 5 mini | OpenAI | 51.7% | $0.04 | 12 |
| 11 | GPT 5.4 mini | OpenAI | 39.2% | $0.03 | 12 |
| 12 | GPT 5.4 | OpenAI | 8.8% | $0.21 | 14 |
| 13 | Claude Opus 4.6 | Anthropic | 0.0% | $1.32 | 185 |