Code Migration
Code · 2026-06-16
Code-migration benchmark evaluating whether language models can reimplement real-world programs in a different programming language while preserving behavior.
Top models (higher is better)
| Model | Score |
|---|---|
| Claude Sonnet 5.5 | 69.8 |
| Gemini 4 Argon | 68.2 |
| GPT-6 Astra | 67.7 |
| GPT-6.1 Sol | 65.1 |
| Muse Wonder | 62.6 |
| Muse Wonder Full | 61.0 |
| Claude Opus 5 | 57.5 |
| GPT-6 Sol | 57.2 |
| GPT-5.6 Sol | 52.9 |
| GPT-5.6 Terra | 47.8 |
| Hy4 Preview | 47.4 |
| Muse Spark 1.3 | 47.4 |
| Opus 4.8 | 47.3 |
| DeepSeek V4.1 Flash | 45.6 |
| GPT-5.5 | 45.2 |
| Grok 4.7 | 44.8 |
| Grok 4.6 | 44.6 |
| Claude Haiku 5.5 | 44.6 |
| GPT-5.6 Luna | 44.5 |
| Sonnet 5 | 44.4 |
| GLM-5.3 | 44.2 |
| DeepSeek V4.1 Flash 1017 | 44.2 |
| Opus 4.7 | 43.9 |
| MiMo-V2.6-Pro | 43.0 |
| GPT-6 Luna | 42.6 |
| DeepSeek V4 Pro 0813 | 41.5 |
| MiMo-V2.6-Flash | 40.9 |
| Sonnet 4.6 | 39.9 |
| Step 5 Preview | 39.0 |
| DeepSeek-V4-Flash-0731 | 38.6 |
| GLM-5.2 | 37.9 |
| Grok 4.5 | 36.6 |
| Gemini 3.8 Flash | 36.5 |
| GPT-5.4 | 35.0 |
| Gemini 3.7 Flash | 34.8 |
| Ember-1 | 32.5 |
| Muse Spark 1.1 | 31.1 |
| Gemini 3.6 Flash | 30.9 |
| Mistral Large 4 Preview | 30.6 |
| Muse Spark 1.2 | 30.0 |