Free (OSS)
Free
Self-hosted with own API key or local models via Ollama/vLLM
Qwen Code
Alibaba's open-source agentic coding CLI, powered by the Qwen3-Coder model family.
Qwen Code is Alibaba's open-source agentic CLI coding tool, built on Google's Gemini CLI architecture and powered by the Qwen3-Coder model family. It runs in the terminal, edits files, executes shell commands, and supports MCP, Skills/SubAgents, daemon mode, and IDE integrations for VS Code, Zed, and JetBrains. The Qwen OAuth free tier was discontinued in April 2026; current options include API keys (Alibaba Cloud, OpenAI, Anthropic, Gemini), Coding Plan subscriptions, third-party providers (OpenRouter, Fireworks AI), and fully self-hosted local setups via Ollama or vLLM under Apache 2.0.
+ Pros
- Qwen3-Coder is fully open-weight and Apache 2.0 — self-host the entire stack on-premises with Ollama or vLLM
- Multi-provider support — use Alibaba Cloud, OpenAI, Anthropic, or Gemini-compatible APIs from a single CLI
- Now ships with IDE integrations for VS Code, Zed, and JetBrains IDEs
- Daemon mode (qwen serve) + TypeScript/Python/Java SDKs enable building custom tools on top of Qwen Code
- MCP and ACP protocol support — extensible with the growing tool ecosystem
- Skills & SubAgents system provides a rich built-in agentic workflow
− Cons
- Qwen OAuth free tier was discontinued in April 2026 — cloud usage now requires an API key or Coding Plan subscription
- Smaller community than Claude Code, Gemini CLI, or OpenHands (24.6k+ GitHub stars)
- Alibaba Cloud API latency can be higher than US-based providers for non-Asia users
- Qwen3-Coder, while strong, still lags behind Claude Opus and GPT-5 on the hardest reasoning tasks
- Daemon mode is still experimental and not recommended for production use
Pricing
Coding Plan
Fixed monthly
Alibaba Cloud Coding Plan — fixed monthly fee, higher quotas via dashscope
Qwen Code is Alibaba's open-source agentic coding CLI, built on Google's Gemini CLI architecture (Apache 2.0) and powered by the Qwen3-Coder model family. The Qwen3-Coder model itself launched in July 2025; the Qwen Code tool followed with its first public release (v0.0.11-nightly.1) on September 9, 2025, and was officially announced in January 2026. It swaps the Gemini model for Alibaba's own Qwen3-Coder family — an open-weight coding model updated through Qwen3.5-Plus and Qwen3.6-Plus releases.
What began as a terminal-native coding agent is now a multi-protocol, multi-platform development tool. Qwen Code supports OpenAI-compatible, Anthropic, and Google GenAI APIs alongside Alibaba Cloud's own endpoints. It ships with IDE integrations for VS Code, Zed, and JetBrains IDEs, and provides TypeScript, Python, and Java SDKs for building custom tools on top of the agent. With 24,624+ GitHub stars, 2,400+ forks, and 456 releases (latest: v0.16.1, May 23, 2026), it is one of the most actively developed open-source coding agents.
Key Capabilities
Terminal-Native Agentic Workflow
Qwen Code runs in the terminal and manages multi-step coding tasks autonomously. It reads and edits local files, executes shell commands, and can navigate complex project structures. The interactive session UI supports slash commands (/help, /clear, /compress, /stats), keyboard shortcuts, and YOLO mode (--yolo) for automatic operations without approval prompts.
Skills & SubAgents
Qwen Code ships with a built-in tool system called Skills and SubAgents, enabling rich agentic workflows. Skills are modular tool definitions that Qwen Code can invoke during a session, while SubAgents allow delegating sub-tasks to specialized agent instances — similar to how Claude Code or Gemini CLI handle complex multi-step reasoning.
MCP and ACP Protocol Support
Qwen Code connects to any MCP (Model Context Protocol) compatible tool or API, giving it access to the growing MCP ecosystem. It also supports ACP (Agent Communication Protocol), enabling inter-agent communication — useful for multi-agent architectures.
File Editing and Shell Execution
The core workflow: Qwen Code reads files in your project, applies changes using the underlying Qwen model, and executes shell commands to verify results. It handles context window management automatically and can reference specific files with @filename syntax in the interactive session.
| Method | Cost | Details |
|---|---|---|
| Self-hosted (Ollama / vLLM) | Free | Run Qwen3 models locally under Apache 2.0 — no API key, no cloud dependency |
| API Key | Pay-as-you-go | Alibaba Cloud ModelStudio, OpenAI, Anthropic, Gemini-compatible endpoints |
| Coding Plan | Fixed monthly | Higher quotas via Alibaba Cloud ModelStudio (Beijing or international) |
| Third-party | Pay-as-you-go | OpenRouter, Fireworks AI |
| Qwen OAuth (discontinued) | Discontinued | Free tier sunset April 15, 2026 |
Version History
Latest release — bug fixes and stability improvements
Qwen3.6-Plus model support, daemon mode refinements, multi-provider auth improvements
Signature Snippet
In the terminal: `qwen` then 'Review all Python files in this data pipeline, identify any SQL injection risks, fix them, add parameterised queries, and write tests for the affected functions.' Qwen Code reads the files, applies fixes using Qwen3.6-Plus, runs the tests, and prints a summary of changes. In headless mode: `qwen -p "Generate unit tests for src/utils.ts"` runs without interactive UI for CI/CD pipelines.More in this Space
Vix
Vix is a Go-native, open-source (AGPL-3.0) AI coding agent that slashes token costs by 40-50% using a stem agent architecture and Tree-sitter virtual filesystem. It rethinks the plan/execute loop — keeping LLM cache warm across Explore, Plan, and Execute phases — while shipping Programmable Workflows, Whiteboard Mode with voice AI, MCP server support, and a self-evolving agent that writes its own scheduled jobs and watchers.
Late — High-Leverage AI Agent Orchestration
Orchestrate an entire AI dev team on 5GB VRAM using ephemeral subagents, exact-match diffs, and a zero-dependency Go binary. Works with any OpenAI-compatible model — local or cloud.
Paca
Paca is a free, open-source, self-hosted Scrum board where AI agents work as equal teammates — assigned to sprints, picking up tasks, and collaborating on BDD specs alongside humans. Built as an alternative to Jira and Linear, it treats AI agents as first-class Scrum members.