Skip to content

Models Benchmarks Providers

Search models and benchmarks /

CountBench

reasoning

Categories: reasoning, spatial reasoning, vision
Modality: image
Language: en
Multilingual: No
Max score: 100
Scoring: %, higher is better
Verified by llm-stats: No

CountBench evaluates object counting capabilities in visual understanding.

Leaderboard

Showing 6 of 6 results

Qwen3.5-27B

97.8%

i
Qwen3.5-35B-A3B

97.8%

i
Qwen3.6-27B

97.8%

i
Qwen3.6 Plus

97.6%

i
Qwen3.5-122B-A10B

97.0%

i
Qwen3 VL 235B A22B Thinking

93.7%

i

Wikibench About Theme Content licensed CC BY-SA 4.0.