Sulat.com / Writing on AIFriday, November 20, 2026
Artificial Intelligence/19 November 2026

Qwen4-Max is free on Nous Portal for five days, served by Inferact on vLLM

Nous Research is giving away five days of free Qwen4-Max on Nous Portal, with Inferact serving it on vLLM. Neither company has published rate limits or an end time.

Glossy purple 3D Qwen logo beside a glass FREE ticket, five calendar tiles and gold tokens flowing from a small server rack.

At a glance

  • Nous Research says Alibaba's Qwen4-Max is free on Nous Portal for five days, in partnership with the Qwen team and Inferact, with native vision and a 1M-token context window.
  • Inferact says the free tokens are served on the Inferact Platform and on vLLM, the open-source inference engine its founders created and maintain.
  • Nous Portal's free plan covers free models only, with standard rate limits and no monthly credits. Nous hasn't published the limits or an end time for the Qwen4-Max window.
  • Portal's catalogue lists a paid Qwen4 Max route at US$1.43 per million input tokens and US$4.28 per million output tokens, against QwenCloud's US$1.50 and US$4.50.

Nous Research has made Alibaba’s Qwen4-Max free on Nous Portal for five days, in partnership with the Qwen team and the inference company Inferact, Nous says on X. The model sits in Portal’s catalogue as qwen/qwen4-max:free.

Nous posted the offer at 8:25am on 19 November, New Zealand time. It describes the model as built for long-horizon work, with native vision and a 1M-token context window. Inferact followed 78 minutes later and said the free tokens are served on the Inferact Platform and on vLLM. Qwen’s own account joined in the same day, asking people to try the model and show what they build.

Chain: Qwen builds Qwen4-Max, Inferact serves it on its platform and vLLM, Nous Portal hands out free tokens for 5 days.
Three parties, one free window: Qwen builds it, Inferact serves it, Nous hands it out.

Portal’s free plan covers free models only, with standard rate limits and no monthly credits. Nous hasn’t published the limits themselves or an end time for the Qwen4-Max window. Five days from the post runs to 23 November.

The same catalogue lists a paid Qwen4 Max route at US$1.43 per million input tokens and US$4.28 per million output tokens. Alibaba’s own QwenCloud list price in Singapore is US$1.50 and US$4.50.

Stat cards: 5 days free, 1M tokens of context, US$0 on the free plan versus US$1.43 in and US$4.28 out on the paid route.
The free route costs nothing; the paid route lists US$1.43 in and US$4.28 out per million tokens.

Inferact was founded by the creators and maintainers of vLLM, the open-source engine for serving language models. Its site says it sells a platform that runs vLLM in a customer’s own cloud or on GPUs it provisions. Nous ran a similar giveaway on 16 November, making Tencent’s Hy4 flash free on Portal for two weeks.

Qwen4-Max itself launched on 16 November, and Alibaba says smaller open-weight Qwen4 models are coming. Nous hasn’t said whether the free route returns once the window closes.


Discover more from Sulat AI & Tech News

Subscribe to get the latest posts sent to your email.