CountBench
Progress Over Time
Interactive timeline showing model performance evolution on CountBench
State-of-the-art frontier
Open
Proprietary
CountBench Leaderboard
7 models
| Context | Cost | License | ||||
|---|---|---|---|---|---|---|
| 1 |  Alibaba Cloud / Qwen Team | 35B | — | — | ||
| 1 | Alibaba Cloud / Qwen Team | 27B | 262K | $0.30 / $2.40 | ||
| 1 | Alibaba Cloud / Qwen Team | 28B | 262K | $0.60 / $3.60 | ||
| 4 | Alibaba Cloud / Qwen Team | — | 1.0M | $0.50 / $3.00 | ||
| 5 | Alibaba Cloud / Qwen Team | 122B | — | — | ||
| 6 | Alibaba Cloud / Qwen Team | 236B | — | — | ||
| 7 | 2B | — | — |
Notice missing or incorrect data?
What is CountBench?
CountBench evaluates object counting capabilities in visual understanding.
CountBench is a image benchmark evaluating models on reasoning, spatial reasoning, and vision tasks. LLM Stats tracks 7 models on this benchmark, scored on a 0–100 scale. The current average is 0.9, with the leader at 1.0.
Compare leaders on the best AI for reasoning, best AI for spatial reasoning and best AI for vision leaderboards.
Current leaders
Qwen3.5-35B-A3B from Alibaba Cloud / Qwen Team currently leads the CountBench leaderboard with a score of 0.978 across 7 evaluated AI models.
FAQ
Common questions about the CountBench benchmark and leaderboard.