Qwen Code Dev Stack
IntermediateAlibaba's open-source coding agent in the terminal, desktop, or browser, running Qwen, GLM, Kimi, or any model you choose.
Published 30 September 2026
About Qwen Code Dev Stack
Qwen Code is Alibaba's open-source, Apache-2.0 coding agent, built by the Qwen team. It began as a fork of Gemini CLI and has grown into its own agent: it reads the repository, plans in Plan Mode, edits files, runs commands and tests, and hands side tasks to subagents. The Qwen models get the deepest integration, and this stack defaults to them, but Qwen Code speaks the OpenAI, Anthropic, and Gemini APIs too, so GLM, Kimi, MiniMax, DeepSeek, Claude, GPT, and local models served by Ollama or vLLM all work, switchable mid-session.
The same agent runs in the terminal, in a desktop app for macOS, Windows, and Linux, in VS Code, Zed, and JetBrains, and in an experimental browser interface. A headless mode and TypeScript, Python, and Java SDKs make it scriptable, and an official GitHub Action lets it review pull requests and triage issues when someone mentions it. Skills, memory, MCP servers, LSP, hooks, git worktrees, and a sandbox that confines commands to a container or macOS Seatbelt extend and fence in what it does.
Most developers run Qwen Code on their own computer, which has to stay on while it works. Its daemon mode shares a running session with a phone browser on the same network through a QR code, including diffs and permission prompts, and chat channels take instructions from Telegram, WeChat, DingTalk, Feishu, and other messaging apps. There are no Alibaba-hosted cloud sessions and no mobile app, though the GitHub Action works on a runner with your machine off. For sessions that run around the clock, the same daemon runs on a VPS or home server as a long-lived service behind an access token, and its browser interface connects to it remotely, though the daemon is still marked experimental.
Code lives on GitHub by default, or on GitLab. A model aggregator, a faster terminal, a session keeper for server use, CI/CD, Docker, and an AI code reviewer are common additions; none of them change how Qwen Code works.
Key Features
- ✓Open-source, Apache-2.0 agent in the terminal, a desktop app, VS Code, Zed, JetBrains, and a browser
- ✓Qwen models by default, plus any OpenAI-, Anthropic-, or Gemini-compatible provider or a local model
- ✓Headless mode, SDKs, and an official GitHub Action for reviews and issue triage
- ✓A daemon that shares sessions with a phone browser and runs on a server behind a token
- ✓Chat channels for Telegram, WeChat, DingTalk, Feishu, and other messaging apps
- ✓Subagents, Plan Mode, skills, memory, MCP, LSP, hooks, and git worktrees
- ✓A sandbox that runs commands in a Docker or Podman container or under macOS Seatbelt
When to Use Qwen Code Dev Stack
- →Running a capable coding agent on Qwen models through an Alibaba Cloud plan or per-token API
- →Using one Alibaba subscription for Qwen, GLM, Kimi, and MiniMax models in the same agent
- →Answering the agent's questions from a phone browser while a long task runs at the desk
- →Triaging issues and reviewing pull requests with the GitHub Action
- →Teams that coordinate in WeChat, DingTalk, or Feishu and want to steer the agent from there
Pros
- The agent is free and open source, with no paid tier or per-seat fee of its own
- Not tied to Qwen: any compatible provider or a local model can be switched in mid-session
- More surfaces than most agents: terminal, desktop app, three editors, browser, and chat apps
- A sandbox and approval modes keep unattended commands contained
- Releases land every few days
Cons
- No free hosted tier since April 2026: Qwen models need an Alibaba Cloud plan or an API key
- The Coding Plan is for interactive use only and has limited slots, so CI and server runs need another way to pay
- No hosted cloud sessions or mobile app; away-from-desk work runs on hardware you keep on
- The browser interface and daemon are still experimental
- Chat channels lean toward Chinese platforms, with Telegram the main option elsewhere
LLM Options for Qwen Code Dev Stack
Qwen Code Dev Stack powered by Qwen
Qwen is the default here and the model family Qwen Code is tuned for, from Qwen-Plus at $0.40 in and $1.20 out per million tokens to the Qwen3.8-Max flagship at $2 and $6. Alibaba's Token Plan and Coding Plan cover it as subscriptions, and most weights are Apache 2.0 for local use.
Qwen Code Dev Stack powered by GLM
Z.ai's GLM is built into Qwen Code's provider list and included in Alibaba's Coding Plan, so one subscription covers it alongside Qwen. On Z.ai directly, the GLM Coding Plan starts at $18 a month, and GLM-5.3 costs $1.40 in and $4.40 out per million tokens.
Qwen Code Dev Stack powered by Kimi
Moonshot's Kimi is a built-in provider and part of Alibaba's Coding Plan. K3 takes the hardest agentic tasks at $3 in and $15 out per million tokens on Moonshot's API, and K2.7-Code handles everyday work at $0.95 and $4.
Qwen Code Dev Stack powered by MiniMax
MiniMax is another built-in provider in Qwen Code and part of Alibaba's Coding Plan. MiniMax-M3 pairs a 1M-token context with $0.30 in and $1.20 out per million tokens, and MiniMax's own Token Plan starts at $22 a month.
Qwen Code Dev Stack powered by DeepSeek
DeepSeek connects as a built-in provider with its own API key. V4.1-Flash is among the cheapest capable coding models, about $0.15 per million input tokens off-peak with a 1M-token context, and V4-Pro takes the harder tasks at $0.66 in off-peak.
Qwen Code Dev Stack powered by Claude
Claude connects through Qwen Code's Anthropic protocol support with an Anthropic API key, billed per token: Sonnet 5 at $2 in and $10 out per million, Opus 5.5 at $4 and $20. It's the pick for the hardest multi-file changes; Claude subscriptions don't apply here.
Qwen Code Dev Stack powered by OpenAI
OpenAI's GPT models connect with an OpenAI API key, billed per token: GPT-5.3-Codex, built for agentic coding, at $1.75 in and $14 out per million, or GPT-6 Sol at $2 and $10. A ChatGPT subscription doesn't cover use in Qwen Code.
These are highlighted picks. To see all the tools, check the LLM category.
Version Control Options for Qwen Code Dev Stack
Qwen Code Dev Stack with GitHub
GitHub is the default here: Qwen Code's official GitHub Action reviews pull requests and triages issues when someone mentions it, and its GitHub channel answers notifications, review requests, and assignments from a running agent.
Qwen Code Dev Stack with GitLab
GitLab bundles source control, merge requests, and CI/CD in one self-hostable platform. Qwen Code's GitLab channel watches your to-dos and answers mentions on issues and merge requests, so the agent can take requests there too.
Qwen Code Dev Stack Add-ons
Each addition below extends this stack with a capability the base stack works fine without. None are required: include the ones your product actually needs when building this stack, and skip the rest.
Server Add-ons
Add a server when you want the agent to keep running somewhere other than your own laptop, reachable at any time and not tied to your machine staying on. Some agents also offer their own managed cloud sessions as an alternative to self-hosting; check the stack's own description for details.
Qwen Code Dev Stack with Hetzner
Hetzner is the default here: its smallest shared plan, CX23, has 4 GB of RAM for about €5.49/mo. Qwen Code documents no memory minimum and the models run at the provider, so 4 GB leaves room for its daemon, builds, and tests. A plain VPS billed by the hour up to a monthly cap.
Qwen Code Dev Stack with DigitalOcean
DigitalOcean's Basic Droplet with 4 GB of RAM and 2 vCPUs costs $24/mo, billed per second up to that cap, with snapshots before a risky unattended run. It has no Qwen Code image, so this is a plain Linux box for the daemon or chat channels.
Qwen Code Dev Stack with Hostinger
Hostinger's KVM 1 has 4 GB of RAM at $6.49/mo introductory, renewing at $11.99/mo, enough for Qwen Code with the models served remotely. A budget pick on a plain Linux image; Hostinger's agent templates target other tools, not Qwen Code.
Qwen Code Dev Stack with Raspberry Pi
A Raspberry Pi 5 is a one-time hardware cost instead of a monthly bill, and Qwen Code installs through npm on 64-bit Linux with Node.js 22. A 4 GB board runs the daemon with the models served remotely, and 8 GB leaves room for builds.
These are highlighted picks. To see all the tools, check the Hosting & Cloud category.
Remote Access Add-ons
Add remote access when you want to reach an agent running on another machine — a VPS or a home server — without exposing it to anyone but you.
Qwen Code Dev Stack with Tailscale
Add Tailscale alongside a server to reach Qwen Code's daemon and browser interface from a laptop or phone anywhere, on a private address. Qwen's own Local Control only covers the same network and warns against exposing the daemon publicly. The personal tier is free.
Session Persistence Add-ons
Add session persistence when the agent runs on a server or over SSH: a terminal multiplexer keeps the session working after the connection drops or the laptop closes, and you reattach from any machine to pick up where it left off.
Qwen Code Dev Stack with tmux
Add tmux when Qwen Code's terminal interface runs on a server: it otherwise ends with the SSH connection, while tmux keeps it working and lets you reattach from any machine. The daemon and chat channels can run as a system service instead.
Qwen Code Dev Stack with Zellij
Add Zellij for the same keep-running sessions with friendlier defaults: shortcuts stay on screen, floating panes suit a test watcher beside Qwen Code, and layouts come back after a reboot, though the running agent doesn't.
Terminal Add-ons
Add a terminal when you want a faster, more configurable place to run the agent than your OS default — most agent CLIs live here all day.
Qwen Code Dev Stack with Ghostty
A GPU-accelerated terminal with native macOS and Linux integration, a solid default for keeping Qwen Code's terminal interface open all day, including the images it can render inline.
Qwen Code Dev Stack with Alacritty
A minimal, GPU-accelerated terminal that does little beyond drawing text fast; pick it when a multiplexer or window manager already handles tabs and splits around Qwen Code.
Qwen Code Dev Stack with WezTerm
A GPU-accelerated terminal with built-in tabs, splits, and multiplexing configured in Lua, able to hold Qwen Code sessions in separate worktrees side by side.
Qwen Code Dev Stack with Kitty
A GPU-accelerated terminal with its own graphics protocol and a scripting layer called kittens, for developers who want deep control over how Qwen Code and its neighbors are laid out.
Qwen Code Dev Stack with iTerm2
The long-standing free macOS terminal, with split panes, profiles, triggers, and tmux integration. The familiar pick for Mac users who want Qwen Code in the terminal they already know.
Code Review Add-ons
Add code review when you want an AI reading every pull request before it merges: it comments on the diff so bugs, security issues, and inconsistencies surface before a human has to catch them.
Qwen Code Dev Stack with CodeRabbit
Add CodeRabbit when you want a second AI reviewer on every pull request Qwen Code opens, separate from the agent's own review: it installs as a GitHub or GitLab app, summarizes the diff, and leaves line-level comments, with a free tier for open-source repos.
Qwen Code Dev Stack with Greptile
Add Greptile when review should read the whole repository rather than just the diff: it indexes the codebase so its GitHub and GitLab comments carry wider context, and it can be self-hosted alongside your own models.
Qwen Code Dev Stack with Cursor Bugbot
Add Cursor Bugbot when a bug-focused reviewer is enough: it reviews GitHub or GitLab pull requests whatever wrote the code, though it bills through a Cursor account and plan rather than its own subscription.
Model Aggregator Add-ons
Add a model aggregator when you want one API key and one bill for models from many providers, with automatic fallback when one of them is down, instead of setting up each provider separately.
Qwen Code Dev Stack with OpenRouter
Add OpenRouter when you want one key and one credit balance for Qwen and hundreds of other models, with fallback when a host is down. Qwen Code lists it as a built-in provider and its own docs suggest it as an alternative to Alibaba's plans; it adds 5.5% on credit purchases.
Qwen Code Dev Stack with LiteLLM
Add LiteLLM when you'd rather run the gateway yourself: an MIT-licensed proxy that puts Qwen, GLM, Kimi, and local models behind one endpoint, with per-key budgets and spend logs, which Qwen Code connects to as a custom provider. Self-hosting it is free.
These are highlighted picks. To see all the tools, check the AI Model Aggregators category.
CI/CD Add-ons
Add CI/CD when you want a dedicated pipeline for running tests, linting, or multi-stage builds before a deploy goes out. Many hosting platforms already redeploy automatically on every push on their own — a CI/CD tool adds the most value on top of that by gating the deploy on a passing test suite, and matters even more when the hosting choice does not auto-deploy at all, such as a self-hosted server.
Qwen Code Dev Stack with GitHub Actions
Add GitHub Actions when every push should run the tests. It's also where Qwen Code's official action runs, reviewing pull requests and triaging issues on a mention, billed to an API key or Token Plan since the Coding Plan excludes automated use.
Qwen Code Dev Stack with GitLab CI/CD
Add GitLab CI/CD when the project is on GitLab and needs automated tests on every push, defined in a .gitlab-ci.yml file that can run entirely on your own runners, with Qwen Code's headless mode as another job if you want it.
These are highlighted picks. To see all the tools, check the CI/CD Pipelines category.
Containerization Add-ons
Add containerization when you want the app packaged the same way across local development, staging, and production, or need to deploy somewhere that isn't a managed serverless platform.
Qwen Code Dev Stack with Docker
Add Docker when Qwen Code's commands should run in a sandbox: its container-based sandbox uses Docker or Podman to keep file changes and shell commands away from the host. Docker also gives the project the same setup locally, on a server, and in CI.
These are highlighted picks. To see all the tools, check the Containerization category.
Frequently Asked Questions about Qwen Code Dev Stack
Can Qwen Code keep working when my computer is off?
Only through its GitHub Action, which reviews pull requests and triages issues on a GitHub runner. Otherwise Qwen Code runs on hardware you keep on: there are no Alibaba-hosted sessions and no mobile app. While your computer stays on, its daemon shares a session with a phone browser on the same network, and chat channels take instructions from Telegram or WeChat. For work that continues with the laptop closed, the daemon runs on a VPS or home server as a long-lived service, reached from a phone browser over a private network such as Tailscale; that mode is still experimental.
Can I use Claude, GPT, or a local model instead of Qwen?
Yes. Qwen Code speaks the OpenAI, Anthropic, and Gemini APIs, so Claude and GPT connect with their own API keys and bill per token; Claude and ChatGPT subscriptions don't apply. DeepSeek, GLM, Kimi, MiniMax, and OpenRouter are built into its provider menu, and Ollama or vLLM serve local models for nothing but hardware. The model can change mid-session, and a subagent can run a different model from the main session, so a cheap model can take side tasks while a stronger one leads.
How does Qwen Code compare to OpenCode?
Both are free, open-source, model-agnostic agents with a terminal interface, a desktop app, editor integrations, and a browser interface. OpenCode has the larger community and OpenCode Go, a $10 flat plan across several labs' open-weight models. Qwen Code is built around Alibaba's models and plans, and goes further on reaching the agent from elsewhere: chat channels for Telegram and Chinese messaging apps, GitHub and GitLab channels, an official GitHub Action, and QR pairing to a phone browser. Its daemon and browser interface are younger than OpenCode's server.
Should I use Alibaba's Coding Plan, Token Plan, or pay per token?
The Coding Plan suits heavy interactive use and is the only one that also covers GLM, Kimi, and MiniMax, but it's a single $50 Pro tier with limited slots restocked daily. The Token Plan's personal editions start far lower and bill in credits, though they're sold for the Singapore region only. Paying per token through Model Studio or another provider suits irregular use, and new Model Studio accounts get a free token allowance to start. The Pricing section has current figures.
Can the Coding Plan power Qwen Code in CI or on a server?
No. Alibaba limits the Coding Plan to interactive use in coding tools and forbids its key in automated scripts or application backends. The GitHub Action, headless runs in CI, and a daemon or chat channel left working unattended on a server all fall on the wrong side of that line. Those need a Token Plan or a per-token API key from Model Studio or another provider, which is one more bill but keeps the subscription safe for your own sessions.
Scores
Popularity2/5
Qwen Code has a sizable open-source following, strongest in China, but far fewer users than Claude Code, Codex, or OpenCode.
Learning Curve3/5
A first task runs within minutes, but choosing between Alibaba's plans and other providers, and between the terminal, desktop, editor, and daemon surfaces, takes more setup than an agent with one plan built in.
Flexibility5/5
Any compatible provider or local model, six surfaces from terminal to chat apps, SDKs in three languages, and skills, MCP, hooks, and a sandbox.
Performance3/5
Qwen models handle agentic coding well and the agent is mature, but results track the chosen model, the daemon is experimental, and nothing runs off hardware you keep on.
Portability5/5
Apache-2.0 licensed, runs on any machine or server, keeps sessions locally, and every model underneath can move to another provider or to your own hardware.
Tools in the Qwen Code Dev Stack Stack
Qwen Code Dev Stack Pricing
Qwen Code is free and open source, so the bill is the models. Alibaba's Token Plan personal editions start at $6/month (limited-time, Singapore region), the Coding Plan Pro is $50/month for interactive use, and per-token API use and other providers bill by usage. Local models cost only hardware. GitHub or GitLab are free for individuals, and an optional server runs about €5.49–24/month.
Apache-2.0 agent for the terminal, desktop, editors, browser, and SDKs; no paid tier of its own.
Token Plan personal from $6/mo (limited-time, regular $8; Singapore region); Coding Plan Pro $50/mo with Qwen, GLM, Kimi, and MiniMax, interactive use only; Model Studio per token (Qwen-Plus $0.40/M input) or any other provider; local models free.
Free tier covers individual devs; GitHub Team is $4/mo, GitLab Premium is $29/mo for more seats.
OpenRouter charges provider prices plus 5.5% on credit purchases; LiteLLM is free to self-host.
Tailscale's personal tier is free for up to 100 devices.
4 GB leaves room for builds next to the agent: about €5.49 at Hetzner, $6.49 intro at Hostinger ($11.99 renewal), $24 at DigitalOcean; or a Raspberry Pi as a one-time cost.