OpenHands Index — leaderboard

Holistic software-engineering agent index from OpenHands, combining issue resolution, multimodal behavior, browsing, and command-line task performance.

Metric: Average Score (%). Source: huggingface.co. 5 models tracked.

Top models

#ModelScore
1OpenHands CodeAct v2.1 (claude-3-5-sonnet-20241022)57.14
2OpenHands CodeAct v2.0 (gpt-4o-2024-11-20)54.2
3AutoCodeRover (gpt-4-turbo-2024-04-09)47.36
4Agentless (gpt-4o-mini-2024-07-18)41.32
5SWE-Agent (claude-3-opus-20240229)38.24

Interactive version: theaggregate.ai/benchmark?slug=openhands-index · How the rankings work · Data refreshed daily, snapshot 2026-07-22.