Download
Qwen3-Coder is the latest and most powerful agentic code model developed by the Qwen team at Alibaba Cloud. Its flagship version, Qwen3-Coder-480B-A35B-Instruct, features a massive 480 billion-parameter Mixture-of-Experts architecture with 35 billion active parameters, delivering top-tier performance on coding and agentic tasks. This model sets new state-of-the-art benchmarks among open models for agentic coding, browser-use, and tool-use, matching performance comparable to leading models like Claude Sonnet. Qwen3-Coder supports an exceptionally long context window of 256,000 tokens, extendable to 1 million tokens using Yarn, enabling repository-scale code understanding and generation. It is capable of handling 358 programming languages, from common to niche, making it versatile for a wide range of development environments. The model integrates a specially designed function call format and supports popular platforms such as Qwen Code and CLINE for agentic coding workflows.
Features
- Large parameter & MoE model: includes a 480B-parameter Mixture-of-Experts model with 35B active parameters in certain variants
- Long context support: native support for up to 256K tokens of context (with ability to extend to ~1M tokens via Yarn) for large files and repository-scale coding tasks
- Wide language support: can generate, understand, and work with 358 coding languages in its recognition and output capabilities
- Instruction and function-call support: includes “Instruct” variants; supports special tokens and tokenizer updates; has function-calling formats enabled through a tool parser
- Fill-in-the-middle (FIM) capability: supports tasks where code is partially given (prefix, suffix, middle) and model fills missing parts appropriately
- Multiple model sizes / quantization: in addition to the huge 480B model, there are 30B-parameter versions; also FP8 quantized variants; documented with example usage and demos
Project Samples
Follow Qwen3-Coder
User Reviews
-
Such a good AI coding model