qwen3-30b-a3b-think is the dedicated thinking variant of Qwen3-30B-A3B MoE model. With 30B total parameters and 3B active parameters and a 128K context window, it permanently enables deep reasoning and is optimized for complex tasks including math, logic and coding. Open-sourced under Apache2.0 license, it is designed for heavy reasoning, scientific analysis and complex code review, without function calling support.