Activereasoningllm Open Source

QwQ-32B

by Alibaba· Released March 2025· Cutoff March 2025

QwQ-32B is a 32-billion-parameter reasoning model developed by Alibaba's Qwen team, designed to excel in complex problem-solving and logical reasoning tasks. It is a distilled version of the larger QwQ-72B model, offering strong performance with reduced computational requirements. The model is open-source and available on Hugging Face.

Official Site 🤗 Hugging Face

Input cost

Free (open source)

Output cost

Free (open source)

Context window

131072 tokens

Max output

8192 tokens

Modalities

text

Parameters

32B

License

Apache-2.0

Capabilities

ReasoningCode GenerationMath Problem SolvingLogical InferenceMultilingual SupportLong Context Understanding

Best For

Complex reasoning tasks, mathematical problem-solving, and code generation where deep logical analysis is required.

Strengths

Strong reasoning capabilities comparable to larger models
Efficient 32B parameter size balances performance and resource usage
Open-source with permissive license
Supports long context up to 128K tokens
Multilingual proficiency

Limitations

Smaller than flagship models like QwQ-72B, may underperform on extremely complex tasks
Not multimodal; text-only input
May require careful prompting for optimal reasoning
Limited to 8K output tokens
Newer model with less community adoption than established alternatives

Use Cases

Advanced mathematical problem solving

Code generation and debugging

Scientific research assistance

Educational tutoring for STEM subjects

Logical puzzle solving

Automated theorem proving

Complex data analysis and reasoning

Improvements Over Previous Model

Distilled from QwQ-72B, achieving similar reasoning performance with fewer parameters
Open-source release under Apache-2.0 license
Supports 128K context window, matching larger models
Optimized for efficient inference on consumer hardware
Improved multilingual reasoning capabilities

Back to all models

Activereasoningllm Open Source

QwQ-32B

by Alibaba· Released March 2025· Cutoff March 2025

Official Site 🤗 Hugging Face

Input cost

Free (open source)

Output cost

Free (open source)

Context window

131072 tokens

Max output

8192 tokens

Modalities

text

Parameters

32B

License

Apache-2.0

Capabilities

ReasoningCode GenerationMath Problem SolvingLogical InferenceMultilingual SupportLong Context Understanding

Best For

Complex reasoning tasks, mathematical problem-solving, and code generation where deep logical analysis is required.

Strengths

Strong reasoning capabilities comparable to larger models
Efficient 32B parameter size balances performance and resource usage
Open-source with permissive license
Supports long context up to 128K tokens
Multilingual proficiency

Limitations

Smaller than flagship models like QwQ-72B, may underperform on extremely complex tasks
Not multimodal; text-only input
May require careful prompting for optimal reasoning
Limited to 8K output tokens
Newer model with less community adoption than established alternatives

Use Cases

Advanced mathematical problem solving

Code generation and debugging

Scientific research assistance

Educational tutoring for STEM subjects

Logical puzzle solving

Automated theorem proving

Complex data analysis and reasoning

Improvements Over Previous Model

Distilled from QwQ-72B, achieving similar reasoning performance with fewer parameters
Open-source release under Apache-2.0 license
Supports 128K context window, matching larger models
Optimized for efficient inference on consumer hardware
Improved multilingual reasoning capabilities

Back to all models