RebusBench (3-Shot): leaderboard
Metric: Exact match: exact match accuracy (%) against the gold answer after lowercasing and stripping whitespace and special characters, on the 1,164 human-authored rebus puzzles of RebusBench, three in-context examples in the text prompt; higher is better. Source: arxiv.org. Saturation forecast: Around September 2028. 9 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 2.5 VL 32B Instruct | 8.08 |
| 2 | Qwen 2.5 VL 7B Instruct | 6.1 |
Interactive version: theaggregate.ai/benchmark?slug=rebusbench-3-shot · How It Works · Data refreshed daily, snapshot 2026-10-07.