HumanEval-Mul (multilingual HumanEval extension) — Distinct multilingual-code extension of HumanEval, not the same benchmark as base HumanEval.
| # | Model | Vendor | Best score | Runs | Last seen |
|---|---|---|---|---|---|
| 1 | DeepSeek-V3 | DeepSeek | 82.6 | 2 | 2026-08-23 |
| 2 | Claude 3.5 Sonnet | Anthropic | 81.7 | 1 | 2026-05-03 |
| 3 | GPT-4o | OpenAI | 80.5 | 1 | 2026-05-03 |
| 4 | DeepSeek-V2.5 | DeepSeek | 77.4 | 2 | 2026-08-23 |
| 5 | Qwen2.5 Instruct 72B | Alibaba | 77.3 | 1 | 2026-05-03 |
| 6 | Llama 3.1 Instruct 405B | Meta | 77.2 | 1 | 2026-05-03 |
| 7 | DeepSeek-V2 | DeepSeek | 69.3 | 1 | 2026-05-03 |
Best tracked score per model (default configuration; source-attributed and verification-tiered). Open a model for its full benchmark surface, provenance and pricing.