qwen-code

Start an interactive session

$ qwen

$ qwen -p "[explain this codebase]"

$ qwen``` then ```@[src/main.ts] [explain this file]

$ qwen``` then ```/auth

$ qwen``` then ```/stats

$ qwen``` then ```/compress

qwen [options]

Qwen Code is an open-source AI coding agent for the terminal from Alibaba, optimized for the Qwen3-Coder model. It helps developers understand large codebases, automate tedious work, and ship faster with agentic coding capabilities.
Built in TypeScript and forked from Gemini CLI, it offers a similar experience but is open to any model supporting the OpenAI completions API. Features include Skills, SubAgents, and Plan Mode for full agentic workflows. The tool supports deep code understanding beyond normal context windows and workflow automation for PR handling, rebases, and formatting.
Authentication options include Qwen OAuth (2,000 free requests/day) or OpenAI-compatible APIs. IDE integrations are available for VS Code, Zed, and JetBrains.

-p PROMPT

Run in headless mode with the provided prompt

/help

Display available commands

Configure authentication method

Clear conversation history

Compress history to save tokens

Show current session information

Submit a bug report

Exit the application

Ctrl+C

Cancel current operation

Exit on empty line

Navigate command history

Reference local files in queries

Requires authentication via Qwen OAuth or OpenAI-compatible API keys. Environment variables OPENAI_API_KEY, OPENAI_BASE_URL, and OPENAI_MODEL can be set for custom providers. Version 0.5.0 supports running four concurrent instances in a single terminal.

Qwen Code was launched by Alibaba in July 2025 alongside the Qwen3-Coder model. The flagship model is Qwen3-Coder-480B-A35B-Instruct, a 480B-parameter Mixture-of-Experts model with 35B active parameters supporting 256K native context and 1M tokens with extrapolation. The tool is released as open-source on GitHub.

gemini(1), claude(1), opencode(1), aider(1)