Loading...
Discovering amazing AI tools


Qwen 3 is the next-generation Qwen series LLM family offering multimodal, agentic, and high-reasoning capabilities across dense and MoE model variants.

Qwen 3 is the next-generation Qwen series LLM family offering multimodal, agentic, and high-reasoning capabilities across dense and MoE model variants.
Qwen3 is the latest generation of the Qwen large-language-model family developed by the Qwen team at Alibaba Cloud. It includes dense and mixture-of-experts (MoE) variants and specialized branches (e.g., Qwen3-VL for vision-language and Qwen3-Coder for code/agentic coding). Qwen3 emphasizes advanced reasoning and instruction-following, a configurable "thinking" mode (enable_thinking) for chain-of-thought style reasoning, native long-context support (32,768 tokens, validated to 131,072 tokens with special techniques), multimodal inputs (image, text, bounding boxes), and first-class support for agentic tool-calls and integrations with tool-call parsers and runtime libraries. The series is published across GitHub and Hugging Face with tooling and CLI integrations (e.g., Qwen Code) to enable local inference, cloud API usage, and developer workflows.



