prahapp
English
Free consultation
Menu

Products ·

Qwen4-Max: Alibaba releases a new flagship model with a 1 million token context

Alibaba has introduced Qwen4-Max, a mixture-of-experts flagship with 2.4 trillion parameters. The company says it beats rivals at coding, but the model weights stay closed.

Alibaba has released Qwen4-Max, the new flagship of its Qwen family. According to the official description it is a mixture-of-experts model with 2.4 trillion parameters and a 1 million token context window. Alibaba says it completes tasks with up to roughly half the tokens of the previous Qwen3.8-Max. The model is not open weight, but the company has announced that it will release weights for smaller Qwen4 models.

Independent numbers come from Artificial Analysis: Qwen4-Max scores 43 on its Intelligence Index and is about 67% cheaper per task than Qwen3.8-Max, though it still trails Kimi K3.1 Pro. According to Bloomberg, Alibaba also says it beats Fable 5.1 and Kimi K3.1 Pro on coding benchmarks. That is a vendor claim, and companies should check it on their own tasks.

There is a safety detail too. South China Morning Post reports that Alibaba filtered the training data to remove offensive cyber capabilities and to reduce chemical, biological, radiological and nuclear risks. The release was among the top stories on Hacker News with about 340 points, so developers are watching it closely.

Is Qwen4-Max available in Europe, and should a company use it? Access through Alibaba's API and apps appears to be open outside China, but we could not confirm exact terms and prices from the primary source. For business data the key question is where the data goes: if the model runs at a Chinese provider, that is a transfer of personal data outside the EU and needs a GDPR assessment. For sensitive data, a model hosted in the EU or on your own infrastructure is the safer choice.

What it means for companies

For Czech companies the lesson is not the leaderboard score. Models change every two to three months, and the best one on paper may not be the best for your documents. That is why we build custom AI agents so the model can be swapped without rewriting the process, and we test it on your own samples first. If you are choosing a model, AI consulting measures quality, cost and data risk side by side. For a comparison of ChatGPT, Gemini, Claude and Copilot see our page on rolling out ChatGPT and other models in a company. Prahapp, the AI division of Buinsoft Technology in Prague, builds a free working prototype within 3 working days.

Sources: Techmeme: Qwen4-Max coverage (Qwen, Bloomberg, SCMP, Artificial Analysis), Qwen blog

Bring us one process or one app idea.

In 30 minutes we tell you whether it is worth handing to an AI agent. Then you get a free working prototype within 3 working days. If you don’t like it, you don’t pay.

  • We reply within 24 hours on working days
  • The engineer who designs it also builds it
  • Write in Czech, English, German, Italian or Turkish

Contact

Write to us

Describe the process or app you want done. We reply within 24 hours on working days, and the prototype within 3 working days is free.