Home / Guides / Qwen Code pricing
Qwen Code pricing: what it costs to run, plan by plan
Qwen Code itself is free: the CLI is open source under Apache-2.0. You pay for the model it calls, in one of three ways. Alibaba Cloud's Coding Plan Pro costs $50 a month for up to 6,000 requests every 5 hours. Qwen OAuth signs in with a Qwen Plus or Pro subscription. Or a Model Studio API key bills per token, from $0.30 per million input tokens and $1.50 per million output on qwen3-coder-flash.
The free daily Qwen OAuth quota people still quote is not how it works now: Qwen's docs say that sign-in needs an active Plus or Pro subscription.
Checked 9 November 2026 against Qwen Code's authentication docs, Alibaba Cloud Coding Plan and Model Studio pricing · By Soren Starck
| Way to pay | Price | Limit |
|---|---|---|
| Qwen Code itself | Free, open source (Apache-2.0) | None: you pay for the model it calls |
| Alibaba Cloud Coding Plan Pro | $50 a month | 6,000 requests per 5 hours, 45,000 a week, 90,000 a month |
| Qwen OAuth (Plus or Pro subscription) | Listed on chat.qwen.ai | Set by the subscription |
| Model Studio API key, pay as you go | From $0.30 input, $1.50 output per million tokens | Per token, no plan cap |
Is Qwen Code free?
The software is. Qwen Code's repository is Apache-2.0 and the CLI costs nothing to install. A free model to go with it is harder to find. Qwen's authentication docs now list Qwen OAuth as "Qwen Account with Subscription" and require an active Plus or Pro subscription from chat.qwen.ai; the release of 5 November 2026 (v0.26.8) added that subscription sign-in to the CLI, the VS Code extension and the desktop app. On the API side, Alibaba's pricing page says its Qwen-Coder models have no free quota in any region. The one free start left: new Model Studio accounts in the Singapore region get 1 million free tokens on some models for 90 days, and qwen3.8-max, the model Alibaba's own Qwen Code setup page uses, is one of them.
Alibaba Cloud Coding Plan: $50 a month
The Coding Plan is a flat monthly subscription for coding agents, sold through Model Studio. Only Pro is on sale: Lite stopped taking new subscribers on 20 March 2026 and stopped renewals on 13 April. Pro costs $50 a month, billed monthly only, and three caps apply at once:
| Window | Coding Plan Pro cap | When it frees |
|---|---|---|
| 5 hours | 6,000 requests | Rolling: each minute frees the usage from 5 hours earlier |
| Week | 45,000 requests | Monday 00:00 (UTC+8), which is Sunday 17:00 in Paris in winter |
| Month | 90,000 requests | 00:00 (UTC+8) on your renewal date |
Whichever cap you reach first stops you. A request here is one model call, not one prompt: Alibaba says a simple task takes about 5 to 10 calls and a complex one 10 to 30 or more, so the 5-hour cap is somewhere between about 200 and 1,200 tasks. When a cap is reached, requests fail until it frees; the plan never falls back to pay as you go.
The plan covers six models: qwen3.7-plus, qwen3.6-plus, qwen3.5-plus, kimi-k2.5, glm-5 and MiniMax-M2.5. It works in Qwen Code and in other coding tools, Claude Code, OpenCode, Cursor, Codex and Cline among them, but only for interactive use: Alibaba's FAQ says calls from scripts, curl or an app backend are refused. Three more terms are worth knowing before you pay: the plan is non-refundable, one account holds one subscription, and new slots are limited and restocked daily at 00:00 UTC+8.
Qwen OAuth with a Plus or Pro subscription
Run /auth in Qwen Code and pick Qwen OAuth to sign in with a chat.qwen.ai account. Usage then counts against that subscription, and /upgrade opens Qwen's plans. Qwen publishes the Plus and Pro prices and quotas on chat.qwen.ai rather than in its docs, and that page did not show its plan list when we read it on 9 November 2026, so this page quotes no figure for them.
Pay per token with an API key
Any Model Studio API key, or a key from another OpenAI-compatible provider, works in Qwen Code with no plan at all. The Qwen-Coder model on Alibaba's price list is qwen3-coder-flash, priced by the length of each request's input. Per million tokens, in the Singapore (international) region:
| qwen3-coder-flash, per million tokens | Input | Output |
|---|---|---|
| Up to 32K input tokens | $0.30 | $1.50 |
| 32K to 128K | $0.50 | $2.50 |
| 128K to 256K | $0.80 | $4.00 |
| 256K to 1M | $1.60 | $9.60 |
The same model costs less in the China (Beijing), Frankfurt and Virginia regions: $0.144 input and $0.574 output per million up to 32K tokens. A long agent session crosses the 32K line quickly, because every turn resends the conversation, so the higher rows are the ones a coding agent pays most of the time. Alibaba's Qwen Code setup page uses qwen3.8-max as its example, at $2 input and $6 output per million in Singapore.
Which is cheaper depends on volume. Pay as you go has no cap and no monthly fee, so light use costs little; the $50 plan wins once you would spend more than that on tokens in a month, and it caps how much you can do in any 5 hours. How Qwen compares with Claude Code as a daily coding agent is in Claude Code vs Qwen.
Checking what you have used
Qwen Code's commands include /stats (also /usage), a dashboard of the tokens your sessions used, and /stats model, which adds an estimated cost per model. Neither reads a plan: they count what this machine sent, not what Alibaba says is left. The Coding Plan balance lives in the Model Studio console, which shows the plan's overall consumption and what remains, with no per-model split. More on both in how to check Qwen usage.
SessionWatcher reads the Coding Plan from your Model Studio console session and keeps its three meters, the 5-hour session, the weekly limit and the monthly limit, each with its reset time, in the Mac menu bar or the Windows taskbar. You see the 5-hour cap closing in before requests start failing. Qwen is in SessionWatcher's Pro plans; trackers that do the same are compared in best Qwen usage trackers.
SessionWatcherSee all three Coding Plan caps
before a request fails.
Native app for the Mac menu bar and the Windows taskbar. Track Claude, Codex and 20 more tools' usage, costs, and rate limits in real time.
“Fast, simple, and does exactly what it should. Definitely worth it.”
@nicojerome on GitHub
macOS 14+. Free for 7 days, no credit card.
Frequently asked questions
How much does Qwen Code cost?
The Qwen Code CLI is free and open source under the Apache-2.0 license. What costs money is the model behind it: Alibaba Cloud's Coding Plan Pro at $50 a month, a Qwen Plus or Pro subscription through Qwen OAuth, or tokens billed per million through a Model Studio API key, from $0.30 input and $1.50 output on qwen3-coder-flash. Read on Qwen's and Alibaba Cloud's pages on 9 November 2026.
Is Qwen Code free?
The tool is, the models mostly are not. Qwen OAuth sign-in now needs an active Plus or Pro subscription from chat.qwen.ai, and Alibaba says its Qwen-Coder models have no free quota in any region. New Model Studio users in the Singapore region get 1 million free tokens on some models, qwen3.8-max among them, for 90 days.
What does the Alibaba Cloud Coding Plan include?
Pro costs $50 a month and allows up to 6,000 requests per 5 hours, 45,000 per week and 90,000 per month, whichever runs out first. It covers qwen3.7-plus, qwen3.6-plus, qwen3.5-plus, kimi-k2.5, glm-5 and MiniMax-M2.5 in coding tools such as Qwen Code, Claude Code, OpenCode and Cline. Lite closed to new subscribers on 20 March 2026.
How many requests does one Qwen Code task use?
Alibaba counts model calls, not your prompts. Its Coding Plan page says a simple task takes about 5 to 10 calls and a complex one 10 to 30 or more, so 6,000 requests per 5 hours comes to somewhere between about 200 and 1,200 tasks, not 6,000.
When does the Coding Plan quota reset?
The 5-hour window rolls: every minute, the usage from 5 hours earlier comes back. The weekly quota resets on Monday at 00:00 UTC+8 and the monthly one at 00:00 UTC+8 on your renewal date. When any of the three is spent, requests stop until it frees; the plan does not switch to pay as you go.
How do I check my Qwen Code usage?
In Qwen Code, /stats (or /usage) opens a dashboard of the tokens your sessions used, and /stats model adds an estimated cost per model. Neither shows what is left of a plan. The Coding Plan balance is in the Alibaba Model Studio console, and SessionWatcher keeps its 5-hour, weekly and monthly meters in the Mac menu bar or the Windows taskbar.
Sources
All read on 9 November 2026:
- Qwen Code on GitHub: the license, sign-in options and the 5 November release notes.
- Qwen Code authentication: Qwen OAuth needs a Plus or Pro subscription; the other ways to connect a model.
- Qwen Code commands: /stats, /usage, /auth and /upgrade.
- Alibaba Cloud Coding Plan: the Pro price, the three caps, resets, supported models and tools, Lite's end.
- Coding Plan FAQ: interactive use only, refunds, where usage shows.
- Model Studio pricing: qwen3-coder-flash by region and input length, free quotas.
- Qwen Code on Model Studio: the qwen3.8-max example and its free tokens.