Epoch AI - Cursorbench: leaderboard

Epoch AI CursorBench evaluates coding agents on realistic Cursor-based software tasks that require editing, debugging, and validating repository changes.

Metric: Score. Source: epoch.ai. 78 models tracked.

Top models

#ModelScoreOverall rank
1claude-opus-5-5_max57.8
2claude-opus-5-5_xhigh56
3claude-opus-5-5_high56
4claude-sonnet-5-5_max55.5
5claude-sonnet-5-5_xhigh53.1
6gemini-4-argon_high52.9
7gemini-4-argon_medium52.8
8claude-opus-5-5_medium52.5
9claude-fable-5-1_max51.8
10claude-fable-5-1_xhigh51.6

Interactive version: theaggregate.ai/benchmark?slug=epoch-ai-cursorbench · How It Works · Data refreshed daily, snapshot 2026-10-20.