Benchmarks / Coding
SWE-bench Multilingual
A 300-instance SWE-bench variant with tasks from 42 repositories across 9 programming languages. Maintained by SWE-bench Team. Score: % Resolved: percentage of task instances solved.
SWE-bench Multilingual
Higher is better · best effort per model
full table below
| # | Model | Effort | Score |
|---|---|---|---|
| 1 | Opus 5 | max | 89.5% |
| 2 | Fable 5.1 | max | 89.1% |
| 3 | Fable 5 | max | 86.6% |
| 4 | Opus 4.8 | max | 84.4% |
| 5 | Qwen 3.8 Flash | max | 81.0% |
| 6 | Opus 4.7 | max | 80.5% |
| 7 | Kimi K2.6 | max | 76.7% |
| 8 | Qwen 3.7 Plus | high | 75.8% |
| 9 | Qwen 3.8 27B | max | 73.8% |
| 10 | Kimi K2.5 | max | 73.0% |
| 11 | Gemini 3 Flash | high | 72.7% |
| 12 | Opus 4.6 | max | 72.0% |
| 13 | MiMo V2 Flash | high | 71.7% |
| 14 | Qwen 3.6 27B | max | 71.3% |
| 15 | Opus 4.5 | high | 70.7% |
| 16 | GLM 5 | high | 69.7% |
| 17 | Qwen 3.5 27B | high | 69.3% |
| 18 | Gemini 3 Pro | high | 68.7% |
| 19 | MiniMax M2.5 | high | 68.3% |
| 20 | Qwen 3.6 35B-A3B | max | 67.2% |
| 21 | Sonnet 4.5 | high | 67.0% |
| 22 | GPT-5.2 | high | 66.7% |
| 23 | GPT-5.2 Codex | xhigh | 66.3% |
| 24 | DeepSeek V3.2 | high | 59.0% |
| 25 | GPT-5 Mini | high | 39.7% |