CaptchaBench - Coordinates: leaderboard
CaptchaBench task (CaptchaMind paper, Wang et al., EMNLP 2026) in which the agent pages through candidate images with arrow buttons to find the one where the animal or fruit sits at the required position. Success rate in percent over 200 generated test episodes, with one ReAct tool-call prompt for every model; higher is better.
Metric: Task success rate (%; 200 test episodes). Source: arxiv.org. Saturation forecast: Around March 2027. 9 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Sonnet 4 | 93.5 | #397 |
| 2 | Claude 3.7 Sonnet | 83.5 | #455 |
| 3 | O3 | 69 | #224 |
| 4 | Gemini 3 Pro | 62 | #135 |
| 5 | Qwen-VL-Max | 55.5 | #1226 |
| 6 | GPT-5 | 35.5 | #163 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=captchabench-coordinates · How It Works · Data refreshed daily, snapshot 2026-11-12.