SlopCodeBench - % Cloned
Code · 2026-01-20
Mean fraction of cloned or duplicated lines, scaled to percent; lower is better.
Top models (lower is better)
| Model | Score |
|---|---|
| GPT-5.5 | 4.7 |
| GPT-5.4 | 5.8 |
| GPT-5.3-Codex | 6.9 |
| GPT-5.4 Mini | 7.6 |
| Opus 4.7 | 8.4 |
| GPT-5.3-Codex-Spark | 8.6 |
| Opus 4.5 | 9.1 |
| Sonnet 4.6 | 9.3 |
| Kimi K2.5 | 9.4 |
| GLM-5.1 | 9.6 |
| GPT-5.2-Codex | 9.7 |
| Opus 4.6 | 10.3 |
| Composer 2 | 10.7 |
| MiniMax M2.7 | 10.8 |
| Kimi K2.6 | 12.9 |