MBPP+ (EvalPlus-augmented) — Raws include MBPP+ matching MBPP-plus exactly; same augmented benchmark, formatting diff only.
| # | Model | Vendor | Best score | Runs | Last seen |
|---|---|---|---|---|---|
| 1 | Kimi K2 Base | Moonshot | 81.3 | 7 | 2026-06-13 |
| 2 | DeepSeek-V3 | DeepSeek | 78.8 | 1 | 2026-06-06 |
| 3 | Hy3 preview-Base | Tencent | 78.7 | 2 | 2026-06-05 |
| 4 | GLM-4.5-Base | Z.ai | 78 | 2 | 2026-06-05 |
| 5 | Qwen2.5 Instruct 72B | Alibaba | 77 | 1 | 2026-06-06 |
| 6 | GPT-4o | OpenAI | 76.2 | 1 | 2026-06-06 |
| 7 | Gemini 2.0 Flash Exp | 75.9 | 1 | 2026-06-06 | |
| 8 | DeepSeek-V3-Base | DeepSeek | 75.5 | 2 | 2026-06-05 |
| 9 | Gemini 1.5 Pro | 75.4 | 1 | 2026-06-06 | |
| 10 | Claude 3.5 Sonnet | Anthropic | 75.1 | 1 | 2026-06-06 |
| 11 | MiMo V2.5 Pro Base | Xiaomi | 74.1 | 3 | 2026-06-05 |
| 12 | Granite 4.1 8B | IBM | 73.8 | 1 | 2026-06-15 |
| 13 | Granite 4.1 30B | IBM | 73.5 | 1 | 2026-06-15 |
| 14 | Llama 3.1 Instruct 405B | Meta | 73 | 1 | 2026-06-06 |
| 15 | Seed Coder 8B Instruct | ByteDance | 72.8 | 1 | 2026-06-05 |
| 16 | DeepSeek-V3.1-Base | DeepSeek | 72.2 | 4 | 2026-06-13 |
| 17 | MiniMax Text 01 | MiniMax | 71.7 | 1 | 2026-06-06 |
| 18 | MiMo V2 Flash Base | Xiaomi | 71.4 | 4 | 2026-06-13 |
| 19 | MiMo V2.5 Base | Xiaomi | 70.9 | 3 | 2026-06-05 |
| 20 | Step3.5 Flash Base | StepFun | 70.6 | 2 | 2026-06-05 |
| 21 | DeepSeek-V3.2-Exp-Base | DeepSeek | 69.8 | 4 | 2026-06-13 |
| 22 | Granite 4.1 30B Base | IBM | 69.6 | 2 | 2026-06-15 |
| 23 | Granite 4.1 3B Base | IBM | 68.3 | 1 | 2026-06-15 |
| 24 | Granite 4.1 8B Base | IBM | 54 | 2 | 2026-06-15 |
Best tracked score per model (default configuration; source-attributed and verification-tiered). Open a model for its full benchmark surface, provenance and pricing.