# Vals Code Migration v1.1 (Code Migration v1.1)

> Reimplementing working programs in another language using the Pi agent harness and revised hidden tests.

Canonical page: https://benchlm.ai/benchmarks/code-migration-v1-1

- Category: [Coding](/coding)
- Last updated: October 20, 2026

Evaluation results published by [Vals AI](https://www.vals.ai/benchmarks/code-migration-v1-1). Published leaderboard aggregates and evaluation settings only. Dataset contents are not included.

## About Code Migration v1.1

- Year: 2026
- Tasks: 120 CLI migration tasks and 10 COBOL tasks
- Format: 75% CLI and 25% COBOL accuracy
- Difficulty: Production code migration
- Paper: [Code Migration v1.1](https://www.vals.ai/benchmarks/code-migration-v1-1)

Version 1.1 replaces mini-SWE-agent with Pi and revises hidden tests. Tasks are unchanged, but scores are not directly comparable with v1. The standalone score combines 120 CLI migration tasks and 10 COBOL tasks at 75% and 25%. Vals Index uses a fixed CLI subset, so its component score differs. Display only.

Code Migration v1.1 is currently displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.

## Leaderboard (28 models)

| Rank | Model | Configuration | Creator | Score |
|------|-------|---------------|---------|-------|
| 1 | [Claude Sonnet 5.5](/models/claude-sonnet-5-5) | — | Anthropic | 86.88% |
| 2 | [Claude Opus 5.5](/models/claude-opus-5-5) | — | Anthropic | 81.01% |
| 3 | [GPT-6 Astra](/models/gpt-6-astra) | max reasoning | OpenAI | 80.58% |
| 4 | [GPT-6.1 Sol](/models/gpt-6-1-sol) | max reasoning | OpenAI | 77.32% |
| 5 | [Muse Wonder](/models/muse-wonder) | max reasoning | Meta | 72.36% |
| 6 | [Gemini 4 Argon](/models/gemini-4-argon) | high reasoning | Google | 70.80% |
| 7 | [DeepSeek V4.1 Flash 1017](/models/deepseek-v4-1-flash) | max reasoning | DeepSeek | 67.52% |
| 8 | [Muse Wonder Full](/models/muse-wonder-full) | max reasoning | Meta | 62.79% |
| 9 | [DeepSeek V4.1 Flash 1017](/models/deepseek-v4-1-flash) | max reasoning | DeepSeek | 58.94% |
| 10 | [Qwen3.8 Max Internal](https://www.vals.ai/models/alibaba_qwen3.8-max-internal) | — | Alibaba | 57.19% |
| 11 | [GLM-5.3](/models/glm-5-3) | max reasoning | Z.AI | 55.07% |
| 12 | [Grok 4.7](/models/grok-4-7) | xhigh reasoning | xAI | 53.22% |
| 13 | [Claude Haiku 5.5](/models/claude-haiku-5-5) | — | Anthropic | 52.36% |
| 14 | [DeepSeek V4 Pro 0813](/models/deepseek-v4-pro-0813) | max reasoning | DeepSeek | 51.99% |
| 15 | [GLM-5.3-Flash](/models/glm-5-3-flash) | max reasoning | Z.AI | 51.65% |
| 16 | [MiMo-V2.6-Pro](/models/mimo-v2-6-pro) | — | Xiaomi | 48.79% |
| 17 | [Step 5 Preview](/models/step-5-preview) | — | StepFun | 47.00% |
| 18 | [GPT-6 Luna](/models/gpt-6-luna) | max reasoning | OpenAI | 46.67% |
| 19 | [Hy4 preview](/models/hy4-preview) | — | Tencent | 45.36% |
| 20 | [MiMo-V2.6-Flash](/models/mimo-v2-6-flash) | — | Xiaomi | 39.55% |
| 21 | [Mercury 3](/models/mercury-3) | high reasoning | Inception | 39.26% |
| 22 | [Gemini 3.8 Flash](/models/gemini-3-8-flash) | high reasoning | Google | 38.65% |
| 23 | [Kimi K3](/models/kimi-k3) | max reasoning | Moonshot AI | 28.59% |
| 24 | [Muse Spark Open](/models/muse-spark-open) | xhigh reasoning | Meta | 26.89% |
| 25 | [Beam 1.0](https://www.vals.ai/models/reflection_Beam-1.0) | — | Reflection | 22.85% |
| 26 | [MiniMax M3](/models/minimax-m3) | — | MiniMax | 20.90% |
| 27 | [Inkling-Small](/models/inkling-small) | 0.99 reasoning | Thinking Machines Lab | 10.43% |
| 28 | [Inkling](/models/inkling) | 0.99 reasoning | Thinking Machines Lab | 8.82% |

## FAQ

### What does Code Migration v1.1 measure?

Reimplementing working programs in another language using the Pi agent harness and revised hidden tests.

### Which model leads the published Code Migration v1.1 snapshot?

Claude Sonnet 5.5 currently leads the published Code Migration v1.1 snapshot with a score of 86.88%.

### How many models are evaluated on Code Migration v1.1?

The October 20, 2026 contains 28 AI models.

### Does Code Migration v1.1 affect BenchLM's overall score?

Not directly. Code Migration v1.1 is still displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.