CaptchaBench - Slide Puzzle: leaderboard

CaptchaBench task (CaptchaMind paper, Wang et al., EMNLP 2026) in which the agent drags a slider piece to the correct position in a single move. 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 2033. 9 models tracked.

Top models

#ModelScoreOverall rank
1Claude 3.7 Sonnet44#455
2Qwen-VL-Max39#1226
3GPT-536#163
4O334#224
5Claude Sonnet 424.5#397

No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.

Interactive version: theaggregate.ai/benchmark?slug=captchabench-slide-puzzle · How It Works · Data refreshed daily, snapshot 2026-11-12.