Multi-SWE-Bench: leaderboard
Multilingual and multi-repository software issue benchmark evaluating whether LLMs can resolve bugs across codebases and languages.
Metric: Score (%). Source: arxiv.org. Source: llm-stats.com. 7 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | MiniMax-M2.7 | 52.7 | #258 |
| 2 | MiniMax-M2.5 | 51.3 | #303 |
| 3 | MiniMax-M2.1 | 49.4 | #289 |
| 4 | Seed 1.8 | 42 | #139 |
| 5 | MiniMax-M2 | 36.2 | #317 |
| 6 | Qwen 3 Coder 480B A35B Instruct | 25.8 | #315 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=multi-swe-bench · How It Works · Data refreshed daily, snapshot 2026-10-20.