Alibaba to Introduce Fees for Heavy Users of Its Upcoming Open-Source AI Model

Chinese technology company Alibaba is preparing to introduce a new monetization strategy for the next version of its Qwen open-source artificial intelligence model by asking large commercial users to share a portion of the revenue they generate from the technology, according to people familiar with the matter. The upcoming Qwen3.8-Max model will be released as an open-source, open-weight AI system, allowing developers to download, modify and deploy it on their own infrastructure. Unlike proprietary models from companies such as OpenAI, Anthropic and Google, Alibabas approach gives developers direct access to the models trained parameters.

Although open-source software is commonly associated with free access, recent developments suggest that major AI firms are redefining how such models generate income. Chinese startup Moonshot recently introduced a similar licensing model for its Kimi K3 AI system, requiring businesses earning more than $20 million annually from commercial services built on the model to negotiate a revenue-sharing agreement.

People familiar with Alibabas plans said the company intends to introduce comparable licensing terms when it launches Qwen3.8-Max, likely as early as next week. The discussions remain private, and the exact revenue-sharing percentage has not yet been finalized. Alibaba has charged developers who access its AI models through Alibaba Cloud while allowing organizations to run many of its open-source models on their own servers without additional licensing fees. The proposed change marks a significant shift in its commercial strategy.

The move reflects a broader trend among Chinese AI companies, which have rapidly gained global attention by releasing highly capable open-source models that compete with leading American AI systems while adopting business models designed to generate revenue from enterprise customers. Industry sources said Moonshot has already signed revenue-sharing agreements with commercial partners, with some deals reportedly requiring as much as 30% of revenue generated from AI-based services. Chinese IT company Chinasoft International has publicly disclosed that it reached a commercial agreement with Moonshot, although financial details were not revealed.

The strategy mirrors the well-established freemium model used by many software companies, where core products remain free while businesses pay for commercial deployment, premium support, optimization services and early access to future releases. DigitalOcean CEO Paddy Srinivasan said companies often pay AI model developers not only for licensing but also for deployment assistance, performance optimization and early access to updated versions. While confirming that DigitalOcean has a commercial agreement with Moonshot, he declined to disclose specific terms.

Pricing remains another competitive advantage for Chinese AI developers. According to publicly available pricing information, Moonshots Kimi K3 is significantly less expensive than comparable proprietary AI offerings such as Anthropics models. Companies providing AI infrastructure continue to earn revenue by improving model efficiency rather than building the models themselves. Together AI Vice President Dan Fu said infrastructure providers focus on optimizing token usage and helping businesses extract greater value from AI applications.

The global open-source AI ecosystem is also expanding beyond China. US-based startup Thinking Machines Lab, founded by former OpenAI Chief Technology Officer Mira Murati, recently introduced its first open-source AI model and is expected to release more advanced versions in the future. Industry executives believe powerful open-source AI models from American developers are likely to become more common, further intensifying competition in the rapidly evolving artificial intelligence market.

Disclaimer: This image is taken from Reuters.