Qwen Code review: five models race, and the winner's code is kept

Not sponsoredAI Agents · Developer Tools

Tested on Linux · Free · Open source

Qwen Code is Alibaba's open-source AI coding agent for the terminal, grown from Gemini CLI. It works with any model provider, and its Agent Arena races several models on the same task and keeps the winner's code.

The good

Agent Arena. We like the Agent Arena: /arena runs up to five models on the same task in separate worktrees, shows a comparison, and applies only the winner. It is still experimental, and its own docs disagree on whether you see the diffs first.
Open source, any model. It is open source under Apache 2.0, works with OpenAI, Anthropic, Gemini and local Ollama providers, and lets you switch between them at runtime.
Many doors. Beyond the terminal, Qwen Code also comes as a desktop app, a browser shell, editor plugins and even chat apps; we tested the terminal.

The not so good

The free sign-in is gone. The free Qwen sign-in ended in April 2026. Alibaba's own Coding Plan now costs 50 dollars a month, sold in limited daily slots.
Setup by hand. Local models are set up by hand: the in-app wizard takes seven steps to type the address, a placeholder key and the model names, though one command line also works. Usage statistics are on by default.

Our verdict

Recommended

Apache 2.0, any model including local ones, and Agent Arena: up to five models race in separate worktrees and you apply the winner. But the free Qwen sign-in ended in April 2026, Alibaba's own plan is 50 dollars a month, and local models are set up by hand.

Reviewed: Qwen Code CLI 0.25.1 · facts checked 2026-10-15

Visit Qwen Code →Alternatives in The Global Almanac

Watch it onYouTube Dailymotion

← All reviews

Stay in the loop

Hear it when it ships

New releases, real benchmarks and the occasional deep-dive. No spam, unsubscribe in one click.

The Brethof network

Everything we build

Voice Pro

Local speech-to-text that learns your voice. Perpetual licence. Our flagship.

PAID · flagship

brethof-brain

Memory for your AI agents — records, rules, the full history and the graph of decisions, already there when a session starts. It answers before you ask.

PAID · free tier