Division
Coding Division
Ranked by SWE-bench Verified: real GitHub issues, resolved end-to-end.
| # | Model | Maker | Division metric | League Score |
|---|---|---|---|---|
| 1 | Claude Opus 5 | Anthropic | 96.0% SWE-bench Verified | 92 |
| 2 | Claude Sonnet 5 | Anthropic | 85.2% SWE-bench Verified | 74 |
| 3 | Kimi K2.6 | Moonshot AI | 80.2% SWE-bench Verified | 70 |
| 4 | Mistral Medium 3.5 | Mistral AI | 77.6% SWE-bench Verified | 59 |
| 5 | Qwen3.5-397B-A17B | Alibaba Qwen | 76.2% SWE-bench Verified | 61 |
| 6 | Gemini 3.1 Pro Preview | Google DeepMind | 75.6% SWE-bench Verified | 52 |
| 7 | GPT-5 | OpenAI | 73.6% SWE-bench Verified | 60 |
| 8 | DeepSeek V4 Pro (0813) | DeepSeek | 73.6% SWE-bench Verified | 63 |
| 9 | Claude Haiku 4.5 | Anthropic | 73.3% SWE-bench Verified | 16 |
| 10 | Nemotron 3 Ultra 550B-A55B | NVIDIA | 71.9% SWE-bench Verified | 57 |
| 11 | Qwen3-Coder-480B-A35B-Instruct | Alibaba Qwen | 69.6% SWE-bench Verified | 41 |
| 12 | MAI-Code-1-Flash | Microsoft AI | 51.6% SWE-bench Verified | 1 |
| 13 | Command A+ | Cohere | 51.0% SWE-bench Verified | 0 |
Division tables regenerate from the model database on every update. Scoring rules: methodology.