Qwen Code
Qwen Code is an open-source coding agent under Apache-2.0 from the Qwen team at Alibaba, originally based on Gemini CLI and developed independently since. It runs in the terminal, in VS Code, Zed and JetBrains, as a desktop app, in a browser and in chat apps, and works with OpenAI, Anthropic, Gemini and Qwen APIs or local models through Ollama and vLLM, switched at runtime, with MCP, a sandbox and subagents.
What does Qwen Code do?
Qwen Code is "the open-source AI coding agent for your terminal, editor, desktop, browser, and chat." Its provider line is the clearest in this category: "Supports OpenAI, Anthropic, Gemini, and Qwen APIs. Any third-party provider or local model (Ollama / vLLM). Switch at runtime." Its README's own pitch to switchers: "If you know Claude Code, you already know Qwen Code."
The feature table claims subagents, agent teams, auto-memory, auto-skills, hooks, MCP, a plan mode, LSP integration, an auto mode, a sandbox, git worktrees, computer use, headless mode and session management. Surfaces are IDE plugins for VS Code, Zed and JetBrains, a desktop app for macOS, Windows and Linux, an experimental web UI started with qwen serve --open, chat channels for Telegram, DingTalk, WeChat and Feishu, SDKs in TypeScript, Python and Java, headless qwen -p, and an experimental daemon mode over HTTP and SSE. An Agent Arena runs several models head to head on one task. The README publishes its own SWE-bench Verified numbers across versions, 76.40 to 77.80 percent average score with the Qwen 3.7 Max model; those are the project's figures. It began as a fork of Gemini CLI v0.8.2.
Key facts
- Licence: Apache-2.0 with no added conditions.
- Providers: OpenAI, Anthropic, Gemini and Qwen APIs, any third-party provider, or local models through Ollama or vLLM, switchable at runtime; /auth configures the provider and key inside a session.
- Surfaces: terminal, VS Code, Zed and JetBrains plugins, a desktop app for macOS, Windows and Linux, an experimental web UI, chat channels (Telegram, DingTalk, WeChat, Feishu), and SDKs in TypeScript, Python and Java.
- Feature table: subagents, agent teams, auto-memory, auto-skills, hooks, MCP, plan mode, LSP integration, auto mode, sandbox, git worktrees, computer use, headless mode, session management.
- Agent Arena runs several models on the same task for comparison.
- The README's own SWE-bench Verified results: 500 cases, 3 trials per version, average scores from 76.40 to 77.80 percent with Qwen 3.7 Max, the project's own measurement.
- Install: a script for Linux and macOS, PowerShell for Windows, npm install -g @qwen-code/qwen-code@latest with Node.js 22 or newer, or brew install qwen-code.
- Originally based on Gemini CLI v0.8.2. Latest tagged release when read: v0.24.0 on 2026-09-16; 27,932 stars on 2026-09-18.
What does it replace, and where does it fall short?
Qwen Code replaces Claude Code for a developer who wants the same terminal workflow with any model, hosted or local, and it says so itself. Its editor plugins and desktop app also cover the agent half of Cursor. On the open-source alternatives to Claude Code page it is the pick for switching between hosted and local models at runtime.
Where it falls short: the web UI and the daemon mode are marked experimental, and the long feature list is a claims table in the README rather than documented behaviour on this page. The benchmark numbers are the project's own. For an agent with a built-in read-only planning mode and a bigger community, OpenCode.
How does Qwen Code run?
Install with the script for Linux and macOS or the PowerShell script for Windows, with npm install -g @qwen-code/qwen-code@latest on Node.js 22 or newer, or with brew install qwen-code. Inside a session, /auth sets the provider and API key, or points at Ollama or vLLM for a local model. The editor plugins, desktop app and SDKs install separately. The README names no paid Qwen Code tier; Alibaba's Model Studio CLI is listed as a separate official tool that extends it.
Who is Qwen Code for?
A developer who wants Claude Code's workflow with the freedom to run Qwen, OpenAI, Anthropic or a local model and to switch mid-session, in the terminal or an editor. Someone who needs only Google's free tier should use Gemini CLI.
What limits does the README state?
From the README: the web UI and the daemon mode are each marked experimental. The feature table and the SWE-bench figures are the project's own claims.
Questions people ask
Is Qwen Code open source?
Yes. The repository is Apache-2.0 with no added conditions. It began as a fork of Google's Gemini CLI at version 0.8.2, which is also Apache-2.0, and its README says it has developed independently since its own version 0.1.
Can Qwen Code run local models?
Yes. The README states support for any third-party provider or local model through Ollama or vLLM, alongside the OpenAI, Anthropic, Gemini and Qwen APIs, and says you can switch at runtime. That makes it one of the terminal agents on this site that can run end to end on your own machines.
Does Qwen Code work inside an editor?
Yes. The README lists plugins for VS Code, Zed and JetBrains, a desktop app for macOS, Windows and Linux, an experimental web UI, and chat channels for Telegram, DingTalk, WeChat and Feishu, in addition to the terminal, headless mode and SDKs in TypeScript, Python and Java.
How does Qwen Code compare with Claude Code?
Its README answers directly: "If you know Claude Code, you already know Qwen Code." The difference is the licence and the models: Apache-2.0 source, and any hosted or local model rather than one vendor's. The open-source alternatives to Claude Code page compares it with OpenCode, Codex CLI, Gemini CLI, Crush, gptme and goose.
How good is Qwen Code on benchmarks?
The README publishes its own SWE-bench Verified evaluation: 500 cases, three trials per version, average scores between 76.40 and 77.80 percent across seven versions with the Qwen 3.7 Max model. Those are the project's measurements, not an independent result, and this site did not re-run them.
Sources
- Qwen Code README and Apache-2.0 LICENSE: github.com/QwenLM/qwen-code, read 2026-09-17.
Compared with the others
On the open-source alternatives to Claude Code page, Qwen Code is the pick for switching between hosted and local models at runtime. Apache-2.0, with OpenAI, Anthropic, Gemini and Qwen APIs or Ollama and vLLM, editor plugins, a desktop app, MCP, a sandbox and subagents. Its README says if you know Claude Code, you already know Qwen Code.
Also on that page: OpenCode for a terminal agent with a planning mode, Crush for the widest provider list with local models, goose for a desktop app and a cli under foundation governance.
On the open-source alternatives to Cursor page, Qwen Code is the pick for terminal plus editor plugins, any model. Apache-2.0, plugins for VS Code, Zed and JetBrains, a desktop app, and OpenAI, Anthropic, Gemini, Qwen or local models through Ollama and vLLM, switched at runtime.
Also on that page: Zed for replacing the editor itself, Kilo Code for autocomplete plus agent, inside vs code or jetbrains, Cline for an agent with reviewable diffs and any model.
More coding assistants and agents
An MIT terminal coding agent with a full-access build agent and a read-only plan agent switched with the Tab key, installers for every platform, a beta desktop app, and the most-starred repository in this directory.
OpenAI's Apache-2.0 coding agent that runs locally in your terminal, with a desktop app and IDE extensions, signed in with a ChatGPT plan or an OpenAI API key, so the code is open and the models are OpenAI's.
Google's Apache-2.0 terminal agent for Gemini models with a free tier of 1,000 requests a day on a personal Google account, built-in search grounding, file and shell tools, MCP support, checkpointing and a GitHub Action.
A high-performance code editor from the creators of Atom and Tree-sitter, GPL-3.0-or-later with Apache-2.0 components, with an agent panel, external agents over ACP, edit prediction, MCP tools, sandboxing, and local models through Ollama or llama.cpp.
An MIT self-hosted control centre, called Agent Canvas, that runs the OpenHands agent or Claude Code, Codex, Gemini and any ACP-compatible agent across local, Docker and cloud backends, with automations triggered by schedules and webhooks.
An Apache-2.0 coding agent for VS Code, JetBrains, the terminal and a desktop app that edits files as reviewable diffs, runs commands with your approval, works with any model provider including Ollama, and connects tools through MCP.
Added September 17, 2026. Every claim above comes from the project's README, LICENSE or model card, read on September 17, 2026, or from the GitHub API on the date shown in the panel. Found an error? Write to reveneau@licheo.com and it is fixed in the next weekly pass. Repository: github.com/QwenLM/qwen-code.
Running one of these inside your own environment, with your own data and your own security rules, is the kind of work Reveneau does. Read how a forward deployed engagement works.