Alibaba to charge major commercial users of next Qwen model

Alibaba plans to require major commercial users of its next Qwen AI model to share revenue generated from the technology, according to two people familiar with the company’s plans.

The move would mark a shift in Alibaba’s approach to its open-weight AI models. While developers would still be able to download and run the model, large commercial users could face new licensing requirements.

Alibaba is expected to introduce the measure alongside the open-weight release of Qwen3.8-Max next week. However, the company has not finalised the revenue-share rate.

The plan follows a similar approach by Chinese AI startup Moonshot. Its Kimi K3 licence requires companies selling the model as a service to reach a commercial agreement after exceeding $20 million in annual revenue.

That agreement can include a revenue share of up to 30%, according to people familiar with Moonshot’s terms.

Until now, Alibaba has charged developers that use its models through Alibaba Cloud. At the same time, most customers have been able to run its open models on their own infrastructure without paying Alibaba.

The change could therefore give Alibaba a new source of revenue from commercial deployments outside its cloud platform.

Alibaba released Qwen3.8-Max earlier this week and described it as its most capable model yet. The model has 2.4 trillion total parameters, with 95 billion active parameters.

Meanwhile, the shift reflects a broader effort by Chinese AI companies to find sustainable business models. They have increasingly released powerful models for free to gain developers and market share.

The strategy resembles a freemium model. Smaller developers can continue using the technology freely, while companies generating significant revenue from it pay for commercial rights or additional services.

DigitalOcean CEO Paddy Srinivasan said his company has already reached a commercial agreement with Moonshot.

“This is a tried and tested open-source ‘freemium’ model,” Srinivasan said.

Dan Fu, vice president of kernels at Together AI, said value also exists in helping customers deploy and optimise AI models.

“At the application layer, there’s value out there for how you use it, how you actually get the models and the tokens to do something useful,” Fu said.

For Alibaba, the proposed change could test whether open AI models can build wide adoption while still generating enough revenue to support their costly development.