bitjaru released bench-v1.0.0 at bitjaru/pixelmind



bitjaru released bench-v1.0.0 of bitjaru/pixelmind · July 24, 2026 22:53
bitjaru / /bitjaru/pixelmindStyleSeed AI Builder Benchmark v1.0
Results
| Agent | Bare | +Rules | +Rules+Gate | Gate vs bare |
|---|---|---|---|---|
| Codex | 74.8 | 76.4 | 80.1 | +5.3 |
| Claude Code | 74.1 | 70.4 | 79.4 | +5.3 |
- 120/120 Claude Code and Codex cells scored; 0 failed.
- Neutral judge: Gemini…
评论
?
参与讨论