Developers wanting a large free coding model with long context can download it, though running 480 billion parameters takes serious hardware.
Kenning's take
What it does
Qwen3-Coder is a family of coding models from the Qwen team, launched in July 2025 with Qwen3-Coder-480B-A35B-Instruct, which has 480 billion parameters with 35 billion active. It supports 256K tokens of context natively and up to 1M with extension, and targets agentic coding, browser use and tool use. Alibaba also released Qwen Code, a command-line coding tool forked from Gemini CLI.
- 480B MoE with 35B active parameters
- 256K context, up to 1M with YaRN
- Trained on 7.5T tokens, 70% code
- Qwen Code command-line tool
- Apache 2.0 open weights
Who it suits. Developers who want an open model for coding agents and IDE tools.
Key facts
- Category
- AI models
- Part of
- Qwen
- Made by
- Alibaba (Qwen team)
- Based in
- China
- Released
- 22 July 2025
- How to get it
- Free weights on Hugging Face and ModelScope; also in Qwen Chat and Alibaba Cloud API
- Works on
- self-hosted, API, CLI
- In the UK
- Yes
- Free plan
- Yes, open weights
FAQ
When was Qwen3-Coder released?
The Qwen team announced it on 22 July 2025.
Is Qwen3-Coder open source?
The weights are released under Apache 2.0.
How big is Qwen3-Coder?
The largest version has 480 billion parameters, 35 billion active.
Facts from the maker's own pages and the sources listed. Checked 2026-10-11.
Similar tools
MAI-UI
Alibaba Tongyi research models that operate smartphone apps by looking at the screen and tapping
Qwen-Image-2.0
Alibaba's 7-billion-parameter model that both creates and edits images, with strong text rendering at 2K
Qwen-Image-Layered
Open Alibaba model that splits an image into separate transparent layers so each part can be edited alone
Qwen-Image-3.0
Alibaba's hosted image model built for text-heavy layouts, taking prompts up to 4,500 tokens