Skip to content

ZeroBench_main with Python

A Python-assisted ZeroBench_main evaluation reported as pass@5.

3 models

Higher is better.

  1. 1DeepSeek V4.1 Flash 1017DeepSeek51.0
  2. 2Qwen3.8 MaxAlibaba49.0
  3. 3Kimi K3Moonshot AI41.0
Higher is better. Bar colour: blue proprietary, orange open weights, green other.

About this benchmark

Metric
Pass@5
Tasks
Visual reasoning questions with Python
Authors
Moonshot AI / ZeroBench authors (2026)

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