ComponentBench (Set-of-Marks): leaderboard
Metric: Pass rate (%; share of tasks whose programmatic verifier passes; shared BrowserGym harness, screenshot with numbered bounding-box overlays, actions address the marks; ComponentBench-Full: 2,910 programmatically verified tasks on 97 canonical UI components across widely used web component libraries; 20-step budget; one deterministic run). Source: arxiv.org. Saturation forecast: Around December 2026. 6 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Flash | 87.1 |
| 2 | GPT-5 Mini | 78.5 |
| 3 | GPT-5.4 | 77 |
| 4 | GPT-5.4 Mini | 74.7 |
| 5 | Gemini 3.1 Flash Lite | 73.5 |
| 6 | Qwen 3 VL 235B A22B | 54.4 |
Interactive version: theaggregate.ai/benchmark?slug=componentbench-set-of-marks · How It Works · Data refreshed daily, snapshot 2026-09-29.