UI-Venus-2 Report - MCA-Bench (1,000-Example Subset): leaderboard
MCA-Bench covers 20 CAPTCHA categories spanning static recognition, point-and-click localization, interactive manipulation and text logic questions. The UI-Venus-2 technical report runs every model on a fixed, category-balanced 1,000-example subset (50 per category) with strict action scoring; Pass@1 in percent, higher is better.
Metric: Pass@1 (%). Source: arxiv.org. Saturation forecast: Around 2031. 7 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3.6 27B | 51.7 | #182 |
| 2 | Kimi K2.6 | 38.7 | #113 |
| 3 | Seed 2.0 Pro | 36.5 | #122 |
| 4 | Qwen 3.5 9B | 30.4 | #391 |
| 5 | Claude Opus 4.6 | 25.9 | #67 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=ui-venus-2-report-mca-bench-1-000-example-subset · How It Works · Data refreshed daily, snapshot 2026-11-05.