KNOWS - Accessibility Tree: leaderboard
KNOWS partial credit (share of evaluation steps passed per task checkpoint, averaged) when a browser agent researches the live web and builds a Google Doc, Sheet or Slide deck, seeing pages only as accessibility trees.
Metric: Average checkpoint fraction (%, overall, BrowserGym-AgentLab, accessibility tree only). Source: alexgill321.github.io. Saturation forecast: Around February 2028. 3 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.5 | 41.9 | #33 |
| 2 | DeepSeek V4 Pro | 31.1 | #112 |
| 3 | Claude Opus 4.7 | 17 | #51 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=knows-accessibility-tree · How It Works · Data refreshed daily, snapshot 2026-11-05.