Best LLM for code migration
- Claude Sonnet 5.5 leads
- 69.8%
- models ranked
- 76
- lead over #2
- +1.7 pts
Code migration (Vals AI) · read October 21
# Code migration (Vals AI), read October 21
leaders = {
"Claude Sonnet 5.5": 69.8%, "Gemini 4 Argon": 68.2%, "GPT-6 Astra": 67.7%,
}
As of October 21, 2026, Anthropic Claude Sonnet 5.5 leads 76 models on Code migration (Vals AI) with 69.8%, ahead of Google Gemini 4 Argon (68.2%) and OpenAI GPT-6 Astra (67.7%).
The full best LLM for coding ranking combines this benchmark with others.
| Rank | Model | Code migration (Vals AI) | Coding evidence | Coding index | Output $/Mtok |
|---|---|---|---|---|---|
| 1 | Claude Sonnet 5.5 Anthropic | 69.8% | 87.3 | $10.00 | |
| 2 | Gemini 4 Argon Google | 68.2% | 81.0 | $10.00 | |
| 3 | GPT-6 Astra OpenAI | 67.7% | 85.1 | $50.00 | |
| 4 | Claude Opus 5.5 Anthropic | 66.7% | 88.8 | $20.00 | |
| 5 | GPT-6.1 Sol OpenAI | 65.1% | 82.4 | $10.00 | |
| 6 | Muse Wonder Meta | 62.5% | 71.9 | $20.00 | |
| 7 | Claude Opus 5 Anthropic | 57.5% | 83.0 | $25.00 | |
| 8 | GPT-6 Sol OpenAI | 57.2% | 76.0 | $10.00 | |
| 9 | Claude Fable 5 Anthropic | 55.1% | 79.3 | $50.00 | |
| 10 | Claude Fable 5.1 Anthropic | 54.6% | 84.2 | $50.00 | |
| 11 | GPT-5.6 Sol OpenAI | 52.9% | 73.5 | $20.00 | |
| 12 | GPT-5.6 Terra OpenAI | 47.8% | 63.9 | $12.00 | |
| 13 | Hy4 preview Tencent | 47.4% | 44.3 | $2.2509 | |
| 14 | Claude Opus 4.8 Anthropic | 47.3% | 65.0 | $25.00 | |
| 15 | DeepSeek V4.1 Flash DeepSeek | 45.6% | 62.5 | $0.78 | |
| 16 | GPT-5.5 OpenAI | 45.2% | 71.3 | $30.00 | |
| 17 | Grok 4.7 xAI | 44.8% | 66.7 | $6.00 | |
| 18 | Claude Haiku 5.5 Anthropic | 44.6% | 70.7 | $0.50 | |
| 19 | Grok 4.6 xAI | 44.6% | 65.7 | $6.00 | |
| 20 | GPT-5.6 Luna OpenAI | 44.5% | 58.0 | $1.20 |
How this list is built
Generated from the dataset on every refresh. Nobody picks the order by hand.
Models by their Code migration (Vals AI) result, as the board publishes it, best first. Vals AI's code migration board, run with one shared harness; its page describes the tasks. The score as Vals AI publishes it.
- OpenRouter, fetched
- Vals AI boards, fetched