Multi-SWE-bench Python (MopenHands): leaderboard

Metric: Resolved rate (%). Source: multi-swe-bench.github.io. Saturation forecast: Estimated already saturated. 12 models tracked.

Top models

#ModelScore
1Claude 3.7 Sonnet (20250219)52.2
2Gemini 2.5 Pro (Preview 03-25)45.8
3Claude 3.5 Sonnet (20241022)39
4DeepSeek V327.8
5DeepSeek R126
6GPT-4o (2024-11-20)25.6
7O3 Mini (2025-01-31) (High)20.4
8O1 (2024-12-17)16
9Llama 4 Maverick14.4
10Doubao-1.5-Pro8.8
11Qwen 2.5 72B Instruct4.4

Interactive version: theaggregate.ai/benchmark?slug=multi-swe-bench-python-mopenhands · How It Works · Data refreshed daily, snapshot 2026-09-26.