Qwen3

Qwen3 is the newest family of open-weight LLMs (0.6B to 235B MoE) from Alibaba. Features switchable "Thinking Mode" for reasoning vs. speed. Strong performance on code/math. Multilingual.

Qwen3 is the latest open-weight large language model family from Alibaba, offering a range of model sizes from 0.6B to 235B parameters, including dense and Mixture-of-Experts (MoE) variants. Its standout feature is the switchable "Thinking Mode," allowing users to toggle between deep reasoning for complex tasks like coding and math, and faster general-purpose chat. The model excels in multilingual support, covering over 100 languages, and demonstrates strong performance in agent-based tasks, creative writing, and instruction following.

Available through platforms like Hugging Face and ModelScope, Qwen3 is designed for flexibility, supporting local deployment on CPU/GPU and integration with frameworks like Transformers, llama.cpp, and vLLM. It builds on previous Qwen iterations with enhanced reasoning, alignment, and tool-use capabilities. The model is licensed under Apache 2.0, encouraging broad adoption and customization.