The short version
Qwen Code is an open-source AI coding agent that runs in your terminal. It comes from the Qwen team at Alibaba, the same group behind the Qwen family of large language models, and it was released to pair with their Qwen3-Coder model. Under the hood it's a fork of Google's Gemini CLI, so the experience feels familiar if you've used that: you type a task, and the agent reads your files, plans, edits code, runs commands, and iterates. What makes Qwen Code distinct is that it's optimized to run on Qwen's own coding models, and it's fully open source, so you can inspect it, modify it, and self-host the pieces you want.
What Qwen Code actually does
Core capabilities
Terminal-native agent
It lives in the command line. You give it a task in plain language and it works across your real files — reading, editing, and running commands to complete multi-step jobs.
Built for Qwen3-Coder
It's tuned to get the most out of Alibaba's Qwen3-Coder models, which are strong open-weight coding models. That pairing is the main reason to reach for it over a generic CLI.
Bring your own endpoint
You can connect it to Qwen's API, Alibaba Cloud Model Studio, or other OpenAI-compatible providers, so you're not locked to a single billing path.
Open source
The code is public (Apache-2.0), a fork of Gemini CLI. You can read exactly what it does, fork it, and extend it — no black box.
Tooling and MCP
Like other modern agents it supports tool use and MCP-style integrations, so it can reach beyond plain text generation into your actual workflow.
Qwen Code vs Claude Code vs Gemini CLI
How the three terminal agents compare
| Aspect | Qwen Code | Claude Code | Gemini CLI |
|---|---|---|---|
| Made by | Alibaba (Qwen team) | Anthropic | |
| Open source | Yes (Apache-2.0) | No | Yes |
| Default model | Qwen3-Coder | Claude models | Gemini models |
| Model choice | Qwen API / Model Studio / OpenAI-compatible | Anthropic-centric | Gemini-centric |
| Interface | Terminal (TUI) | Terminal (TUI) | Terminal (TUI) |
| Best for | Teams already on Qwen models, open-weight fans | Anthropic subscribers who want a capable agent | Google/Gemini users |
All three are terminal-first. Plans, pricing, and model support change over time — check each project's official site for current details.
How much does Qwen Code cost?
Qwen Code itself is free and open source — there's no license fee to run it. What you pay for is the model behind it. If you point it at Qwen's API or Alibaba Cloud Model Studio, you pay for model usage (pay-as-you-go or a coding/token plan, depending on region and account). If you run a Qwen open-weight model locally, the tool costs nothing and there's no per-token bill, just your own hardware. As always, the sticker is the model, not the agent.
If you'd rather not live in the terminal
Qwen Code is terminal-first, and for plenty of developers that's exactly right. But if you want the same idea — a local, bring-your-own-model coding agent — with an actual desktop window, a visible file tree, a scrollable chat pane, and model switching you click instead of configure, you want a graphical agent. Wuwei is a free, open-source (MIT) one: it runs locally on Windows, macOS, and Linux, reads your real files, and lets you pick any model from a list — including OpenAI-compatible endpoints and local models — in a normal GUI. You can bring your own key, run a local model, or try its free hosted models with no login. It's a different shape from Qwen Code, and it costs nothing to see if the GUI workflow fits you better.
FAQ
Is Qwen Code free?
The Qwen Code tool is free and open source (Apache-2.0). Running it is only as expensive as the model you connect it to: pay-as-you-go via Qwen's API or Alibaba Cloud Model Studio, or free if you run a Qwen open-weight model locally on your own hardware.
Is Qwen Code the same as Gemini CLI?
Qwen Code is a fork of Google's Gemini CLI, so the user experience is very similar. The main difference is that Qwen Code is tuned to run best on Alibaba's Qwen3-Coder models rather than Gemini models.
What model does Qwen Code use?
It's built around the Qwen3-Coder models from Alibaba, but because it can talk to OpenAI-compatible endpoints you can also point it at other providers or a local model.
Does Qwen Code have a GUI?
No. Qwen Code is a terminal-based agent (a TUI). If you want the same bring-your-own-model idea in a real desktop window with a file tree and clickable model switching, a graphical agent like Wuwei is a free, open-source alternative.
Qwen Code vs Claude Code — which is better?
Neither is simply better. Qwen Code is open source and built for Qwen3-Coder models; Claude Code is closed and built around Anthropic's Claude models. Pick based on which models you want to run and whether open source matters to you. Both are terminal agents.