SlopCodeBench - % AST-Grep
Code · 2026-01-20
Mean fraction of AST-Grep-flagged lines, scaled to percent; lower is better.
Top models (lower is better)
| Model | Score |
|---|---|
| MiniMax M2.7 | 22.7 |
| GPT-5.4 | 24.0 |
| GPT-5.5 | 24.9 |
| Opus 4.6 | 28.8 |
| Opus 4.5 | 29.4 |
| Kimi K2.5 | 29.7 |
| Sonnet 4.6 | 29.8 |
| GLM-5.1 | 30.1 |
| GPT-5.4 Mini | 30.5 |
| GPT-5.3-Codex | 31.4 |
| Composer 2 | 31.8 |
| Opus 4.7 | 32.7 |
| GPT-5.3-Codex-Spark | 34.0 |
| Kimi K2.6 | 35.9 |
| GPT-5.2-Codex | 36.4 |