Multi-SWE-Bench (java) — leaderboard
Metric: Score (%). Source: huggingface.co. 41 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | CodeArts Agent + CodeArts-MiniMax-M2.5 | 47.86 |
| 2 | InfCode + GPT5.2 | 39.06 |
| 3 | iSWE-Agent | 33.86 |
| 4 | iSWE-OpenModels | 31.25 |
| 5 | MSWE-agent + Gemini-2.5-Pro | 28.91 |
| 6 | MSWE-agent + CodeArts-MiniMax-M2.5 | 28 |
| 7 | MSWE-agent + Claude-3.7-Sonnet | 23.44 |
| 8 | MagentLess + DeepSeek-R1 | 22.66 |
| 9 | MopenHands + Claude-3.7-Sonnet | 21.88 |
| 10 | MagentLess + Gemini-2.5-Pro | 21.88 |
Interactive version: theaggregate.ai/benchmark?slug=multi-swe-bench-java · How the rankings work · Data refreshed daily, snapshot 2026-07-22.