Qwen pricing: free app, API from $0.15 per million tokens, $50 Coding Plan

In brief

Qwen pricing starts at $0, because the Qwen app is free, and the API bills per million tokens: Qwen3.8 Flash costs $0.15 in and $0.47 out, Qwen3.8 Max $2 and $6, as of November 2026. The one flat subscription for developers, Alibaba's Coding Plan, costs $50 a month.

How much does Qwen cost?

Qwen costs nothing in its own chat app and a few cents to a few dollars per million tokens through Alibaba's API. Developers can also pay a flat $50 a month for the Coding Plan. We read every price below on 14 November 2026 from Alibaba Cloud's Model Studio price list, which says it was last updated on 12 November, and from its Coding Plan page. The pages we read don't say whether tax is added at checkout.

Four violet glass cubes rising in height on black plinths in a charcoal studio, with a brass balance scale and a small chrome hourglass beside the shortest cube.
Way to pay for Qwen, read 14 November 2026PriceWhat you getWho it suits
Qwen app (chat.qwen.ai)$0Chat, no price list publishedEveryday questions
API, Qwen3.8 Flash$0.15 in, $0.47 out per 1M tokensCheapest current Qwen3.8 modelHigh volume, simple jobs
API, Qwen3.7 Plus$0.40 in, $1.60 out per 1M tokens, up to 256K inputThe mid tierMost apps
API, Qwen3.8 Max$2 in, $6 out per 1M tokensThe top Qwen modelHard reasoning
Coding Plan Pro$50 a monthUp to 90,000 requests a monthCoding agents and editors
Token Plan (China Beijing only)39 to 312 yuan a month20,000 to 200,000 creditsDevelopers in mainland China
Qwen3.8 Max costs about 13 times Flash, on input and on output
inputoutput$0$2$4$6$8US dollars per million tokens, Singapore list, read 14 Nov 2026Qwen3.7 Max$2.50$7.50the older top modelQwen3.8 Max$2$6the top modelQwen3.8 27B$0.50$3open weightsQwen3.7 Plus$0.40$1.60up to 256K inputQwen3.8 Flash$0.15$0.47cheapest Qwen3.8Qwen3.7 Flash$0.03$0.13up to 32K input

What does the Qwen API cost per million tokens?

The Qwen API costs from $0.03 to $2.50 per million input tokens and from $0.13 to $7.50 per million output tokens on Alibaba's Singapore list. That's the international price. Alibaba also lists a separate China (Beijing) price, and for Qwen3.8 Max, the one model we read it for, it runs lower. Here are the current models, read 14 November 2026:

ModelInput per 1MOutput per 1MFree quota (Singapore)
qwen3.8-max$2$61M tokens
qwen3.7-max$2.50$7.501M tokens
qwen3.8-2.4t-a95b$2$61M tokens
qwen3.7-plus, up to 256K input$0.40$1.601M tokens
qwen3.7-plus, 256K to 1M input$1.20$4.801M tokens
qwen3.8-27b$0.50$31M tokens
qwen3.8-flash$0.15$0.471M tokens
qwen3.7-flash, up to 32K input$0.03$0.131M tokens
qwen-flash, up to 256K input$0.05$0.40None

Three catches sit under that table. Alibaba labels the Qwen3.7 Plus price a limited-time 20% discount off the list. The older qwen-max line is $1.60 in and $6.40 out, with no free quota. And Beijing lists Qwen3.8 Max at $1.65 in and $4.951 out, about 17% under Singapore. Alibaba's page says batch calls cost 50% of real-time prices for models that support them. It bills cached tokens at their own rates, which its context cache page sets at 10% of the input price for explicit hits on most models. That page says Qwen3.8 Max is an exception, and it prints no figure for it, so budget Max at the full price.

What does one real job cost on each Qwen model?

A job of 1,000 requests costs $11.00 on Qwen3.8 Max and 84 cents on Qwen3.8 Flash, so Max charges 13 times as much for the same traffic. The per-million prices are hard to feel, so we priced out one job for you. Take 1,000 requests, each sending 4,000 tokens in and getting 500 back. That's 4 million input tokens and half a million output tokens. We used no cache and no batch discount, so these are ceilings.

Model, Singapore list read 14 November 2026Input costOutput cost1,000 requests
Qwen3.7 Max$10.00$3.75$13.75
Qwen3.8 Max$8.00$3.00$11.00
Qwen3.8 27B$2.00$1.50$3.50
Qwen3.7 Plus$1.60$0.80$2.40
Qwen3.8 Flash$0.60$0.235$0.835
Qwen flash, older line$0.20$0.20$0.40
Qwen3.7 Flash, short prompts$0.12$0.065$0.185

Output is the smaller part of the bill here, because each of the replies is short. If you ask for long answers or run thinking models, the output tokens grow fast, and the ranking holds but the gaps change. Pick the model by what your task needs. A Flash model costs a few cents for a thousand simple replies, and Max costs eleven dollars for the same count.

Does Qwen have a paid plan for the chat app?

Qwen publishes no price for its chat app outside China, as of November 2026. The app at chat.qwen.ai is free, and the page we could read shows no plan list or limits. Our Qwen vs ChatGPT page covers the app's terms, data storage and who it suits.

China is different. TechNode reported on 11 August 2026 that the Qwen app added paid office assistant plans at 19, 49 and 128 yuan a month, or 200, 568 and 1,499 yuan a year. It said basic access "remains free" and gave no quota numbers. Separate credit packs pay for video generation. We found no sign that these plans exist outside of China.

If you hit a cap or want one place for several models, a flat plan can cost less than juggling accounts. Whizi's free plan gives you 10 messages a day on Whizi AI, Auto and Claude Haiku 5.5, plus 2 pictures a month and 5 photo uploads a day, with no card. Pro costs $29.99 a month for 2,500 credits and carries Claude, GPT and Gemini, and its first week is $1.99. Pro also has the Qwen3.7 Plus model at 1 credit a message. It isn't Qwen's own app, so Qwen's terms don't apply to it.

How much is the Qwen Coding Plan?

The Qwen Coding Plan costs $50 a month, and Pro is the only tier left to buy. Alibaba's Coding Plan page, last updated 4 November 2026, shows three caps that apply together: "Up to 6,000 requests per 5 hours", "Up to 45,000 requests per week" and "Up to 90,000 requests per month". Alibaba applies whichever cap you reach first. Lite stopped taking new buyers on 20 March 2026, and its renewals and upgrades ended on 13 April.

Pro is on limited availability. Alibaba says that slots go first come, first served and that they restock daily, so you may not get one on your first try. The plan covers Qwen3.7 Plus, Qwen3.6 Plus, Kimi K2.5, GLM-5 and MiniMax-M2.5. It doesn't list Qwen3.8 Max.

The plan is cheap if you use most of it. At the full 90,000 requests, $50 works out to $0.00056 a request. Run those 90,000 through the Qwen3.7 Plus API at our 4,000-in, 500-out size and you'd pay $216. The real gap depends on your request size, and a coding agent may send several requests for one task. For people who stay well under the cap, pay-as-you-go can cost less than $50.

Qwen planPriceCapsWhere
Coding Plan Pro, read 14 Nov 2026$50 a month6,000 per 5 hours, 45,000 a week, 90,000 a monthModel Studio, international
Coding Plan LiteClosed to new buyersEnded 20 March 2026Model Studio, international
Token Plan Lite39 yuan a month20,000 credits, 1 to 2 agentsChina (Beijing) only
Token Plan Essential79 yuan a month45,000 credits, 2 to 3 agentsChina (Beijing) only
Token Plan Standard139 yuan a month80,000 credits, 3 to 4 agentsChina (Beijing) only
Token Plan Pro312 yuan a month200,000 credits, 6 to 8 agentsChina (Beijing) only

The Token Plan prices come from Alibaba's Token Plan page, which says the personal edition is "currently available only in the China (Beijing) region". New buyers get half off Lite and Essential until 23:59:59 on 30 November 2026, Beijing time. From 20:00 on 9 November, Beijing time, the page says all four quotas rose and Pro's price fell. The page's own table now shows Pro at 312 yuan and 200,000 credits. ChainCatcher reported that Pro fell from 499 yuan and from 180,000 credits. Existing Pro subscribers keep their old price until the next billing cycle.

Is there a free Qwen API tier?

Yes, but only in Singapore: each current model gets 1 million free tokens for 90 days. Alibaba's price list says the free quota runs from the latest of activation, model release or approval, and "Failed requests are not billed and do not consume your free quota". Beijing and other regions get none. Our Qwen vs ChatGPT page covers the same quota from another angle.

One million tokens sounds like a lot, and it isn't. At our job size of 4,500 tokens a request, it covers about 222 requests on one model. The quota is per model, so if you try Max, Plus and Flash, you get three of the allowances. After 90 days, or once a model's million runs out, you pay the list price. The older qwen-plus, qwen-max and qwen-flash lines have no free quota at all. Whether the app itself is free to use at work is a different question, which is Qwen free answers.

Which Qwen option should you pay for?

Use the free app if you only chat, Flash if you build something that sends many short requests, and the Coding Plan if an editor or agent will use most of its 90,000 monthly requests. Pay for Max only when Flash and Plus fail on your own test. It costs 13 times Flash for the same traffic.

Check the region before you commit to any of them. The Singapore list is what you pay from most countries, and Beijing's cheaper price and Token Plan are for mainland China accounts. And watch the dates: the Plus discount is limited-time, and the Token Plan's new-user price ends on 30 November 2026. For how Qwen's quality compares to what you'd pay for elsewhere, Qwen vs DeepSeek and DeepSeek pricing are the next two reads.

Workflow checklist
  • Start in the free Qwen app if you only need chat, and skip the API until you build something
  • Price your own job in tokens before choosing a model: 1,000 requests of 4,000 in and 500 out cost $0.84 on Flash and $11.00 on Max
  • Use the free 1 million tokens per model in Singapore to test, and note that it expires after 90 days
  • Check whether your account is Singapore or Beijing before you read any price, because the two lists differ
  • Buy Coding Plan Pro at $50 only if you expect to use a large share of its 90,000 monthly requests
  • Test Flash and Plus on your real prompts first, and move to Max only when they fail
Common questions

Frequently asked questions

How much does Qwen cost per month?

Nothing in the Qwen app, and $50 a month for Alibaba's Coding Plan Pro, as of November 2026. The API has no monthly fee: you pay per million tokens, from $0.15 in and $0.47 out on Qwen3.8 Flash to $2 and $6 on Qwen3.8 Max. Kimi pricing shows what a rival Chinese lab charges.

Is Qwen free?

The Qwen chat app is free and publishes no price list, and the API gives 1 million free tokens per model for 90 days in Singapore only, as of November 2026. After that it bills at list price. Check the app's terms before you use it for work, as we covered in Qwen vs ChatGPT.

Is Qwen cheaper than ChatGPT?

On the API, often yes. Qwen3.8 Max costs $2 in and $6 out per million tokens, against $2 and $10 for GPT-6.1 Sol on OpenAI's pricing page as read on 14 November 2026. But per finished task, Qwen vs ChatGPT found Qwen3.8 Max cost $5.41 and GPT-6.1 Sol $0.72.

How much does the Qwen Coding Plan cost?

The Coding Plan costs $50 a month on Pro, which is the only tier open to new buyers as of November 2026. It caps you at 6,000 requests per 5 hours, 45,000 a week and 90,000 a month. Lite is closed. Slots are limited and restock daily.

Can you run Qwen for free on your own computer?

Yes for the open-weight models, and you pay for the hardware. Artificial Analysis lists Qwen3.8 27B under Apache 2.0, as we read on 12 October 2026 for Qwen vs ChatGPT. Qwen3.8 Max is closed, so it only runs through Alibaba's API at the prices above.