Code Migration — Code Quality
Code · 2026-06-16
Code-quality rubric score (0–5) sub-track of the Code Migration benchmark. Split out because it is a 0–5 rubric, not a 0–100 pass rate.
Top models (higher is better)
| Model | Score |
|---|---|
| Sonnet 5 | 4.9 |
| Claude Haiku 5.5 | 4.8 |
| Claude Opus 5 | 4.8 |
| Opus 4.8 | 4.7 |
| Claude Sonnet 5.5 | 4.6 |
| Opus 4.7 | 4.6 |
| GLM-5.3 | 4.5 |
| GPT-6.1 Sol | 4.5 |
| GPT-6 Astra | 4.5 |
| Hy4 Preview | 4.4 |
| DeepSeek V4.1 Flash | 4.4 |
| Step 5 Preview | 4.4 |
| GPT-6 Sol | 4.4 |
| GLM-5.3 Flash | 4.4 |
| Muse Spark 1.3 | 4.3 |
| Qwen3.8-Max | 4.3 |
| Muse Wonder Full | 4.2 |
| MiMo-V2.6-Pro | 4.2 |
| Qwen3.8 27B | 4.2 |
| Mistral Large 4 Preview | 4.1 |
| Grok 4.7 | 4.1 |
| MiMo-V2.6-Flash | 4.1 |
| Grok 4.5 | 4.1 |
| Grok 4.6 | 4.1 |
| Muse Wonder | 4.1 |
| DeepSeek V4 Pro 0813 | 4.0 |
| Sonnet 4.6 | 4.0 |
| DeepSeek V4.1 Flash 1017 | 4.0 |
| GLM-5.2 | 4.0 |
| Ember-1 | 4.0 |
| DeepSeek-V4-Flash-0731 | 3.9 |
| GPT-5.6 Terra | 3.9 |
| Kimi K3 | 3.9 |
| GPT-5.6 Sol | 3.8 |
| GPT-6 Luna | 3.8 |
| Gemini 3.8 Flash | 3.7 |
| MiniMax M3 | 3.7 |
| Gemini 3.7 Flash | 3.7 |
| GPT-5.4 Nano | 3.7 |
| GPT-5.4 | 3.6 |