Vals AI Code Migration: leaderboard
Vals AI Code Migration — large-scale code migration tasks (framework/language version upgrades) run agentically; accuracy on private migration task set.
Metric: Accuracy (%). Source: www.vals.ai. Saturation forecast: Around January 2027. 79 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Sonnet 5.5 | 69.83 | #11 |
| 2 | Gemini 4 Argon | 68.17 | #8 |
| 3 | GPT-6 Astra | 67.74 | #6 |
| 4 | Claude Opus 5.5 | 66.65 | #3 |
| 5 | GPT-6.1 Sol | 65.12 | #7 |
| 6 | Muse Wonder | 62.55 | #15 |
| 7 | Muse Wonder Full | 61.03 | #10 |
| 8 | Claude Opus 5 | 57.47 | #12 |
| 9 | GPT-6 Sol | 57.2 | #14 |
| 10 | Claude Fable 5 | 55.06 | #16 |
| 11 | Claude Fable 5.1 | 54.61 | #9 |
| 12 | GPT-5.6 Sol | 52.92 | #19 |
| 13 | GPT-5.6 Terra | 47.8 | #41 |
| 14 | Hy4 preview | 47.43 | #38 |
| 15 | Muse Spark 1.3 (Max) | 47.41 | #24 (Muse Spark 1.3) |
Interactive version: theaggregate.ai/benchmark?slug=vals-ai-code-migration · How It Works · Data refreshed daily, snapshot 2026-10-20.