Qwen’s flagship mixture-of-experts model with 235B total parameters and ~22B activated per token for efficient high-capacity inference. It features a dedicated “thinking” mode optimized for mathematics, coding, science, and multi-step logic tasks, along with large context support. The model delivers strong benchmark performance among open-weight reasoning models, making it well-suited for research copilots, complex workflows, and systems that prioritize structured, high-quality reasoning.
