Skip to content

CountBench

A visual counting benchmark that tests whether a model can count objects and entities reliably in complex scenes.

8 models

Higher is better.

  1. 1Qwen3.6-27BAlibaba97.8
  2. 2Gemini 3 ProGoogle97.3
  3. 3Qwen3.5 397BAlibaba97.2
  4. 4Kimi K2.5Moonshot AI94.1
  5. 5GPT-5.2OpenAI91.9
  6. 6Claude Opus 4.5Anthropic90.6
  7. 7LFM2.5-VL-3BLiquid AI87.3
  8. 8LFM2.5-VL-450MLiquid AI73.3
Higher is better. Bar colour: blue proprietary, orange open weights, green other.

About this benchmark

Metric
Image-grounded counting
Tasks
Visual counting tasks
Authors
Qwen (2026)

Scores are as reported by model developers or evaluators and may use different settings (for example reasoning effort or tools). Compare with care.