SlopCodeBench - Isolated Solved
Code · 2026-01-20
SlopCodeBench evaluates code at a sequence of development checkpoints. Isolated Solved is the percentage of checkpoints whose isolated tests pass.
Top models (higher is better)
| Model | Score |
|---|---|
| GPT-5.5 | 28.1 |
| GPT-5.3-Codex | 26.0 |
| GPT-5.4 | 23.5 |
| GPT-5.2-Codex | 21.9 |
| Opus 4.6 | 20.9 |
| Opus 4.7 | 20.9 |
| Kimi K2.6 | 18.9 |
| Opus 4.5 | 17.3 |
| Sonnet 4.6 | 16.8 |
| Composer 2 | 16.3 |
| GLM-5.1 | 13.8 |
| GPT-5.4 Mini | 13.8 |
| Kimi K2.5 | 9.7 |
| GPT-5.3-Codex-Spark | 8.2 |
| MiniMax M2.7 | 4.1 |