Alibaba is preparing to change how large commercial users access its upcoming open-source AI model, with the Chinese technology company planning to seek a share of revenue generated by major users, according to 2 people familiar with the plans.
Alibaba’s upcoming Qwen3.8-Max is expected to follow the open-source, open-weight approach, allowing developers to download the underlying learned settings and run or adapt the model. In contrast, US companies such as OpenAI, Anthropic and Google operate closed-source models.
The planned approach is similar to Moonshot’s Kimi K3 licensing terms. While open-source AI is often viewed as free, Kimi K3 requires companies offering the model as a service and generating more than $20 million in annual sales to enter a commercial agreement with Moonshot.
Alibaba is expected to introduce a similar provision next week, the 2 sources said. They requested anonymity because the plans are not public. Alibaba currently charges developers when its models are hosted on its cloud platform, while most open-source models can be used in customers’ own data centres without payment.
The move highlights how Chinese AI companies are developing commercial models around open-source systems as they compete with US rivals.
Revenue-sharing deals take shape
Moonshot can reportedly seek up to a 30% revenue share under Kimi K3 licensing terms. Alibaba also plans to seek a revenue share, although the rate has not been decided. Chinasoft International recently disclosed a revenue-sharing agreement with Moonshot but did not reveal the percentage.
US firms are also working with Chinese AI models. “You pay for collaboration with these open-weight model labs to make sure that you’re optimizing your deployment. You pay for getting early access for the next revision of the model,” said Paddy Srinivasan, CEO of DigitalOcean Holdings. He confirmed a commercial agreement with Moonshot but did not disclose details.
“This is a tried and tested open-source ‘freemium’ model,” Srinivasan said.
Kimi K3 costs about 1/3 of Anthropic’s Fable model based on listed input and output token prices.
Dan Fu of Together AI said AI software companies earn through service optimisation. “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.
Meanwhile, US-based Thinking Machines Lab has released its 1st open-source model and is expected to launch more powerful models.
“I don’t see a fundamental barrier” to powerful open-source US models, said Fireworks AI CEO Lin Qiao. “We are really waiting for that to happen.”
Also read: Viksit Workforce for a Viksit Bharat
Do Follow: The Mainstream LinkedIn | The Mainstream Facebook | The Mainstream Youtube | The Mainstream Twitter
About us:
The Mainstream is a premier platform delivering the latest updates and informed perspectives across the technology business and cyber landscape. Built on research-driven, thought leadership and original intellectual property, The Mainstream also curates summits & conferences that convene decision makers to explore how technology reshapes industries and leadership. With a growing presence in India and globally across the Middle East, Africa, ASEAN, the USA, the UK and Australia, The Mainstream carries a vision to bring the latest happenings and insights to 8.2 billion people and to place technology at the centre of conversation for leaders navigating the future.