Multi-SWE-bench mini (MopenHands): leaderboard
Multi-SWE-bench mini, a 400-issue subset of Multi-SWE-bench (ByteDance Seed, 2025), whose agents resolve real GitHub issues by editing the repository; a fix counts only when the project's tests pass. Scored as the percentage of issues resolved under MopenHands, the benchmark team's multilingual port of OpenHands; higher is better.
Metric: Resolved rate (%). Source: multi-swe-bench.github.io. Saturation forecast: Around December 2027. 13 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3 Coder 480B A35B | 25.75 | #677 |
| 2 | Claude 3.7 Sonnet (20250219) | 18.25 | #399 |
| 3 | Gemini 2.5 Pro (Preview 03-25) | 18 | #234 |
| 4 | Claude 3.5 Sonnet (20241022) | 12.25 | #635 |
| 5 | DeepSeek V3 | 7.5 | #705 |
| 6 | DeepSeek R1 | 7 | #467 |
| 7 | GPT-4o (2024-11-20) | 6.75 | #864 |
| 8 | O3 Mini (2025-01-31) (High) | 5.75 | #486 (O3 Mini (2025-01-31)) |
| 9 | Llama 4 Maverick | 5.5 | #1105 |
| 10 | O1 (2024-12-17) | 4.75 | #310 |
| 11 | Qwen 2.5 72B Instruct | 2.25 | #1093 |
| 12 | Doubao-1.5-Pro | 2 | #585 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=multi-swe-bench-mini-mopenhands · How It Works · Data refreshed daily, snapshot 2026-11-12.