Requests limits overview

[squaretus]

What would you like to be added?

I'm currently using the free plan.
As far as I know, it allows for 1,000 requests per day, which should be enough for full-fledged work in my case.
However, I see that without completing even one task today, I've reached my limit with the following message:

  ✕ Qwen OAuth quota exceeded: Your free daily quota has been reached.

    To continue using Qwen Code without waiting, upgrade to the Alibaba Cloud Coding Plan:
      China:       https://help.aliyun.com/zh/model-studio/coding-plan
      Global/Intl: https://www.alibabacloud.com/help/en/model-studio/coding-plan

    After subscribing, run /auth to configure your Coding Plan API key.

It would be great for UX metrics to see how much of my quota is left, so I can understand whether it's sufficient or not, and decide whether to upgrade to a paid plan, for example.

One of the key killer features I would also make the ability to view the list of requests, so I can see what managed to consume 1,000 requests in an hour of work today.

Why is this needed?

Это необходимо в первую очередь для принятия решений о переходе на новые лимиты и платные тарифы

так же, при дебаге запросов к квену, можно было бы увеличить эффектичность, запретив некоторые вызовы для агента, чтобы не тратить запросы там, где модель начинает зацикливаться например

Additional context

I am attaching the debug information for today's session.

Qwen Code: 0.14.0 (de66ee1)
Среда выполнения: Node.js v25.9.0 / npm 11.11.0
ОС: darwin arm64 (25.3.0)
Аутентификация: Qwen OAuth
Модель: coder-model
ID сессии: 83d24b3d-2757-469f-a936-c12bdf7ad578
Песочница: no sandbox
Прокси: no proxy
Использование памяти: 367.9 MB

ask me if you need additional info

[YoshiaKefasu]

@squaretus

[squaretus]

@squaretus

YoshiaKefasu

I saw this after I received the limit hammer.
In any case, I don't think this changes the relevance of the feature I proposed.

Thanks for the quick response!

[wiilwaalka]

What about using it through the command line without connecting it to other IDEs? I’ve been doing some great work creating CLI tools with Qwen Coder and exploring related projects on Termux. I haven't extensively used other options, but I've experimented with a few. I came across this news today and that really worried me. I'm not sure where to turn now. Please don't treat us poor dudes like this, Alibaba!

[YoshiaKefasu]

Nvidia's CEO Just Said the Quiet Part Out Loud: “Intelligence is about to become a commodity.”

[tanzhenxin]

Related: #3203

[lianzhiweivincent-cloud]

If the real blocker now is not Qwen Code itself but the gap between free and large paid plan, one practical route is to keep the Qwen Code workflow and only change the endpoint underneath it.

I just wrote down the small-budget version of that logic here:
https://black-eagle-ai.vercel.app/guides/qwen-code-after-free-tier/

The short version is:

• keep the IDE workflow
• switch to an OpenAI-compatible endpoint you can actually buy today
• start with the smallest paid test first instead of committing to a big monthly plan

At current public rates, a $5 starter test is roughly enough for about 3.8M input + 3.8M output tokens on economy text, or about 1.5M + 1.5M on reasoning text, which is usually enough to verify whether the workflow is worth keeping.

Disclosure: I am involved with Black Eagle AI, so please ignore the link if you only want official Qwen options. The main idea still applies either way: preserve the workflow first, then decide whether the larger paid route is worth it.

[lianzhiweivincent-cloud]

@wiilwaalka if you mainly need the CLI workflow and just want a place to turn right now, I would keep the Qwen Code setup and swap only the endpoint underneath it. I wrote the small-budget version here: https://black-eagle-ai.vercel.app/guides/qwen-code-after-free-tier/ and a tiny first-test flow here: https://black-eagle-ai.vercel.app/starter-trial/ . Same base_url pattern, no workflow rewrite.

[lianzhiweivincent-cloud]

@wiilwaalka if you mainly want the CLI workflow and just need somewhere to turn right now, the shortest path I’ve found is to keep the same setup and swap only the endpoint underneath it.

• small-budget guide: https://black-eagle-ai.vercel.app/guides/qwen-code-after-free-tier/
• tiny first test: https://black-eagle-ai.vercel.app/starter-trial/

Same base_url pattern, no workflow rewrite. If you want the narrowest setup, reply starter and I’ll keep it to one account, one model, one request.