Benchmark
SWE-bench Multilingual
reasoning
code
text
多语言
A multilingual benchmark for issue resolving in software engineering that covers Java, TypeScript, JavaScript, Go, Rust, C, and C++. Contains 1,632 high-quality instances carefully annotated from 2,456 candidates by 68 expert annotators, designed to evaluate Large Language Models across diverse software ecosystems beyond Python.
语言EN
满分1
参评模型14
模型排名
| 名次 | 模型 | 机构 | 分数 | 来源 |
|---|---|---|---|---|
| 1 | Ornith 1.0 397B | DeepReinforce | 78.9 | 来源 ↗ |
| 2 | Qwen3.7 Max | Alibaba Cloud / Qwen Team | 78.3 | 来源 ↗ |
| 3 | Claude Sonnet 5 | Anthropic | 78.3 | 来源 ↗ |
| 4 | Grok 4.5 | xAI | 78.0 | 来源 ↗ |
| 5 | LongCat-2.0 | Meituan | 77.3 | 来源 ↗ |
| 6 | Ornith 1.0 35B | DeepReinforce | 69.3 | 来源 ↗ |
| 7 | Nemotron 3 Ultra 550B A55B | NVIDIA | 67.7 | 来源 ↗ |
| 8 | Laguna XS 2.1 | Poolside | 63.1 | 来源 ↗ |
| 9 | DeepSeek-V3.2-Exp | DeepSeek | 57.9 | 来源 ↗ |
| 10 | DeepSeek-V3.1 | DeepSeek | 54.5 | 来源 ↗ |
| 11 | Ornith 1.0 9B | DeepReinforce | 52.0 | 来源 ↗ |
| 12 | Kimi K2 Instruct | Moonshot AI | 47.3 | 来源 ↗ |
| 13 | Kimi K2-Instruct-0905 | Moonshot AI | 47.3 | 来源 ↗ |
| 14 | DeepSeek-R1-0528 | DeepSeek | 30.5 | 来源 ↗ |