Vals Code Migration v1.1 (Code Migration v1.1)
We show this table for reference; we do not rank on it.
Reimplementing working programs in another language using the Pi agent harness and revised hidden tests.
Evaluation results published by Vals AI. Published leaderboard aggregates and evaluation settings only. Dataset contents are not included.
Code Migration v1.1 score on Code Migration v1.1 — October 20, 2026
We mirror the published code migration v1.1 score view for Code Migration v1.1. Claude Sonnet 5.5 leads the public snapshot at 86.88%, followed by Claude Opus 5.5 (81.01%) and GPT-6 Astra (80.58%). We do not use these results to rank models overall.
Claude Sonnet 5.5
Anthropic
Claude Opus 5.5
Anthropic
GPT-6 Astra
OpenAI
max reasoning
28 modelsCodingCurrentDisplay onlyUpdated October 20, 2026
Code Migration v1.1 score table (28 models)
| Rank | Model / configuration | Score | Parameters (B) | Open / closed |
|---|---|---|---|---|
| 1 | Claude Sonnet 5.5Anthropic | 86.88% | Not reported | Closed |
| 2 | Claude Opus 5.5Anthropic | 81.01% | Not reported | Closed |
| 3 | GPT-6 AstraOpenAImax reasoning | 80.58% | Not reported | Closed |
| 4 | GPT-6.1 SolOpenAImax reasoning | 77.32% | Not reported | Closed |
| 5 | Muse WonderMetamax reasoning | 72.36% | Not reported | Closed |
| 6 | Gemini 4 ArgonGooglehigh reasoning | 70.80% | Not reported | Closed |
| 7 | DeepSeek V4.1 Flash 1017DeepSeekmax reasoning | 67.52% | Not reported | Open |
| 8 | Muse Wonder FullMetamax reasoning | 62.79% | Not reported | Closed |
| 9 | DeepSeek V4.1 Flash 1017DeepSeekmax reasoning | 58.94% | Not reported | Open |
| 10 | Qwen3.8 Max InternalAlibaba | 57.19% | Not reported | Unknown |
| 11 | GLM-5.3Z.AImax reasoning | 55.07% | Not reported | Open |
| 12 | Grok 4.7xAIxhigh reasoning | 53.22% | Not reported | Closed |
| 13 | Claude Haiku 5.5Anthropic | 52.36% | Not reported | Closed |
| 14 | DeepSeek V4 Pro 0813DeepSeekmax reasoning | 51.99% | Not reported | Open |
| 15 | GLM-5.3-FlashZ.AImax reasoning | 51.65% | Not reported | Open |
| 16 | MiMo-V2.6-ProXiaomi | 48.79% | Not reported | Open |
| 17 | Step 5 PreviewStepFun | 47.00% | Not reported | Open |
| 18 | GPT-6 LunaOpenAImax reasoning | 46.67% | Not reported | Closed |
| 19 | Hy4 previewTencent | 45.36% | Not reported | Open |
| 20 | MiMo-V2.6-FlashXiaomi | 39.55% | Not reported | Open |
| 21 | Mercury 3Inceptionhigh reasoning | 39.26% | Not reported | Closed |
| 22 | Gemini 3.8 FlashGooglehigh reasoning | 38.65% | Not reported | Closed |
| 23 | Kimi K3Moonshot AImax reasoning | 28.59% | Not reported | Open |
| 24 | Muse Spark OpenMetaxhigh reasoning | 26.89% | Not reported | Open |
| 25 | Beam 1.0Reflection | 22.85% | Not reported | Unknown |
| 26 | MiniMax M3MiniMax | 20.90% | Not reported | Open |
| 27 | Inkling-SmallThinking Machines Lab0.99 reasoning | 10.43% | Not reported | Open |
| 28 | InklingThinking Machines Lab0.99 reasoning | 8.82% | Not reported | Open |
How Code Migration v1.1 is shown here
BenchLM mirrors the public Vals AI Code Migration v1.1 leaderboard captured from https://www.vals.ai/benchmarks/code-migration-v1-1 and updated by Vals on October 20, 2026. The snapshot preserves overall scores, uncertainty, latency, cost-per-test metadata, and task-level scores where Vals publishes them.
Code Migration v1.1 is display only on BenchLM. Vals proprietary or Vals-hosted aggregate views are useful context, but BenchLM does not use them as weighted ranking inputs or as a replacement for benchmark-native source records.
Snapshot
The published Code Migration v1.1 snapshot places Claude Sonnet 5.5 first at 86.88%. The third row is 6.30 points behind. The broader top-10 range is 29.69 points, so the table still separates the published systems.
28 models have been evaluated on Code Migration v1.1. The benchmark falls in the Coding category. We keep external benchmark mirrors separate from the weighted global scoring system, so these results remain source-specific evidence. Code Migration v1.1 is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.
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
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.
Freshness and provenance
Version
Code Migration v1.1 2026
Refresh cadence
Quarterly
Staleness state
Current
Question availability
Public benchmark set
BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.
Questions
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 86.88% code migration v1.1 score. BenchLM shows this benchmark for display only and does not use it in overall rankings.
How many models are evaluated on Code Migration v1.1?
The October 20, 2026 snapshot contains 28 AI models.
Know when it’s worth switching models
The model to choose, the cheaper alternative, and the release we would wait on.
Read a sample issueJoin 6,100+ subscribers.
One email each week. Unsubscribe anytime.