Qwen-Code: The Coding Agent That Lives in the Digital World

Developers deserve more than autocomplete they deserve an agent that ships. Qwen‑Code isn’t just another CLI; it’s a coding companion that reasons, plans, and executes end‑to‑end from the terminal, so teams ship faster and fix smarter.

Forked from Gemini CLI and optimised for Qwen3‑Coder, it translates messy debug sessions into AI‑powered clarity, navigates multi‑file repos and automates the boring work so engineering focuses on impact.

With free OAuth access 2,000 requests/day and 60RPM, any developer can sign in and start building in under a minute. It’s open, extensible and designed for agentic workflows, code understanding, tool‑use and planning baked in. From junior to genius, this is how command lines become creative lines.

Core paradigm

• Traditional Coding CLI/IDEs: Static commands, manual orchestration
• Qwen-Code: Agentic coding: plans, tools, multi-step workflows

Context handling

• Traditional Coding CLI/IDEs: Limited file-by-file
• Qwen-Code: Optimised for Qwen3-Coder with large context and repo-wide reasoning

Setup speed

• Traditional Coding CLI/IDEs: Tokens, providers, plugins
• Qwen-Code: One-command OAuth; 2,000 requests/day free, 60RPM

Autonomy

• Traditional Coding CLI/IDEs: Suggestions over actions
• Qwen-Code: Executes tasks, navigates repos, uses tools via function calls

Cost control

• Traditional Coding CLI/IDEs: Token micromanagement
• Qwen-Code: Free OAuth tier with automatic credential caching

Extensibility

• Traditional Coding CLI/IDEs: Plugin overhead
• Qwen-Code: Forked from Gemini CLI with customised prompts and tool protocols

Learning curve

• Traditional Coding CLI/IDEs: Complex automation
• Qwen-Code: “qwen” and go; clear OpenAI-compatible env support

Reliability

• Traditional Coding CLI/IDEs: Manual retries
• Qwen-Code: Model fallback to maintain service quality when needed

• Free usage at scale: Qwen OAuth offers 2,000 requests/day and 60 requests/minute; sign in from the CLI, and credentials are managed automatically.
• Release velocity: Nightly and stable releases are active, including v0.0.5‑nightly series and v0.0.6 updates.
• Ecosystem credibility: Qwen3‑Coder 480B‑A35B‑Instruct supports 256K context (1M via extrapolation) and is positioned for agentic coding and tool use, precisely what Qwen‑Code unlocks at the CLI.
• Community traction: Developers highlight strong real‑world results using Qwen‑Code with local and hosted models, reflecting grassroots adoption.

Core

• Research‑purpose CLI adapted from Gemini CLI; customised prompts and function‑calling protocol for Qwen‑Coder

Languages/models

• Works via OpenAI‑compatible API; examples show OPENAI_MODEL=qwen3‑coder‑plus

Authentication

• Qwen OAuth (recommended), OpenAI‑compatible environment variables, automatic credential caching

Free tier

• 2,000 requests/day, 60 requests/min via Qwen OAuth

Features

• Code understanding & editing across large repos; workflow automation; enhanced parser optimised for Qwen‑Coder

Install

• npm i -g @qwen-code/qwen-code (or install from source)

Runtime

• Node.js 20+ required

Config

• OPENAI_API_KEY, OPENAI_BASE_URL, OPENAI_MODEL; .env supported

Input (Prompt): “Refactor this repository to extract a reusable auth module and add tests.”

• Agentic Response (Expected): Plan steps, inspect project files, propose module boundaries, edit files, draft tests, guided by Qwen‑Coder protocols.

Input (Prompt): “Diagnose why our Node service spikes CPU under load and patch it.”

• Agentic Response (Expected): Inspect code paths, identify hotspots, propose fixes, apply changes across files with repo‑wide reasoning.

Input (Prompt): “Migrate this REST route to serverless and update infra config.”

• Agentic Response (Expected): Generate migration plan, edit handlers, update configs and validate build steps with minimal manual orchestration.

• GitHub: Full source, docs and issues on the official repository.
• Model and ecosystem: Qwen3‑Coder announcement and docs for installation and OpenAI‑compatible configuration.
• Community examples: Real‑world feedback and usage patterns from developer forums can be seen at Reddit.

Prerequisites

• Node.js 20+ installed.

Install via npm

• npm i -g @qwen-code/qwen-code
• qwen --version

Authenticate (Recommended)

• Run: qwen
• Sign in via browser with Qwen OAuth; credentials are cached automatically.
• Free tier: 2,000 requests/day, 60RPM.

Configure (OpenAI‑compatible mode)

• Export or .env:
• export OPENAI_API_KEY="your_api_key_here"
• export OPENAI_BASE_URL="your_api_endpoint"
• export OPENAI_MODEL="your_model_choice"
• Then: Run qwen

Version: v0.0.7-nightly.0

• Date: Pre-release
• Highlights: Stability and UX polish across the CLI: fixes for terminal flicker during login, improved OpenAI-compatible content generation, refined init behaviour and end-to-end test reliability geared to smoother day-to-day agent workflows.

Version: v0.0.6

• Date: Latest stable
• Highlights: Major integration wave: Qwen OAuth added for seamless sign-in, OpenRouter support, sandbox/docker fixes, telemetry service, usage logging for Qwen integration, token usage optimisations and default model/fallback behaviour tightened.

Upcoming Capabilities

• Continued optimisation of parser/tool protocols for Qwen‑Coder agentic tasks and repo‑scale operations.

Q1. What is Qwen‑Code?

An AI‑powered command‑line workflow tool adapted from Gemini CLI and optimised for Qwen‑Coder, enabling planning, tool‑use and multi‑file code workflows directly in the terminal.

Q2. Is there a free tier?

Yes, Qwen OAuth offers 2,000 requests/day and 60 requests/minute with automatic credential management and refresh.

Q3. How do I install it?

npm i -g @qwen-code/qwen-code; Node.js 20+ required. Alternatively, build from source via the GitHub repo.

Q4. What can it automate beyond codegen?

Repo analysis, refactoring, test generation, git automation, changelogs and file operations powering multi‑step workflows.

Q5. Where do I see releases and contribute?

Visit the GitHub Releases page for version history and submit issues/PRs on the repo.

Try Qwen‑Code Today. Install in minutes. Build with momentum. From code chaos to command‑line clarity.

Visit official GitHub 

Start command: qwen with OAuth for free 2,000 requests/day and 60RPM

Install via npm: npm i -g @qwen-code/qwen-code

Configuration guide: Official Doc Website 

Code smarter. Debug with drive. Level up from junior to senior dev faster, with an agent that actually ships.

Need help with AI transformation? Partner with OneClick to unlock your AI potential. Get in touch today!