Skip to content

QwQ

An experimental Qwen research model for deep reasoning

Visit official site

What QwQ is

The model is described as an experimental research model from the Qwen team, named Qwen with Questions and focused on reasoning ability. The material says it approaches questions through self-questioning and reflection.

What you can do with it

  • Explore reasoning through self-questioning and reflection
  • Work on mathematics problems
  • Work on programming tasks
  • Provide a complex problem and observe the reasoning process

Benchmark figures

  • GPQA is listed at 65.2%
  • AIME is listed at 50.0%
  • LiveCodeBench is listed at 50.0%

What to watch out for

  • The model may switch between languages unexpectedly
  • Recursive reasoning loops may produce long, inconclusive answers
  • Enhanced safety measures are described as needed
  • Common-sense reasoning and language understanding are listed as areas to improve

Pros & cons

✓ What we like

  • The material describes mathematics and programming as its areas of focus
  • The approach is described as self-questioning and reflection

! What to watch out for

  • The model may switch between languages unexpectedly
  • Recursive reasoning loops may produce long, inconclusive answers

FAQ

What is QwQ?

The material describes an experimental research model from the Qwen team, Qwen with Questions, focused on reasoning.

Which figures are listed?

The material lists 65.2% on GPQA, 50.0% on AIME and 50.0% on LiveCodeBench.

Last reviewed: 2026-11-15

More LLM API platform tools

View all →

How we review