QwQ-32B is Alibaba’s open-source dense reasoning-specialized LLM in the Qwen family with 32.5B parameters and a 128K context window. Deep thinking reasoning is enabled by default. Trained with reinforcement learning, it delivers strong performance on complex reasoning tasks in mathematics, science and coding. It supports tool calling and is open-sourced under Apache2.0 license, suitable for math problem-solving, scientific analysis, complex coding and heavy long-chain reasoning tasks.