Chinese AI players, from big ones like Baidu to smaller ones like Manus AI, are increasingly turning to open-source licensing models for their projects, CNBC reports.
Image source: Solen Feyissa/unsplash.com
The trend toward open-sourcing AI models was sparked by Chinese startup DeepSeek, which released its R1 reasoning model earlier this year, challenging American dominance in the AI market and sending shares of leading tech companies tumbling. It also raised questions among investors about the huge costs companies spend on developing large language models, as DeepSeek has succeeded with far less.
According to several analysts, the most significant impact of DeepSeek’s success on the market has been the acceleration of open-source AI model adoption. “DeepSeek’s success proves that open-source strategies can lead to faster innovation and wider adoption,” said Wei Sun of Counterpoint Research, noting that many organizations have adopted DeepSeek’s model.
On March 16, Baidu unveiled a new version of its Ernie 4.5 base AI model, as well as a new reasoning model, Ernie X1, making them free for individual users. Baidu also plans to open-source a series of Ernie 4.5 models in late June. Experts noted that Baidu’s move to open source represents a broader shift in China’s AI market, away from a business strategy that focuses on proprietary software licensing.
Lian Jye Su, chief analyst at technology research and advisory group Omdia, pointed out that Baidu has always stuck to a proprietary business model and opposed open source, but companies like DeepSeek have proven that open source models can be just as competitive and reliable as proprietary ones.
DeepSeek R1 is licensed under the MIT license, which Counterpoint describes as one of the most accessible and widely used open source licenses, allowing unrestricted use, modification, and distribution of the software, including for commercial purposes.
While DeepSeek’s AI model itself is free, the startup charges for the use of an API that allows the models and their capabilities to be integrated into other companies’ apps. However, its price tag is touted as being much more affordable than the latest offerings from OpenAI and Anthropic, whose models are considered “closed” because their datasets and algorithms are not publicly available.
In addition to Baidu, other Chinese tech giants have also begun to offer open-source models. Alibaba Cloud said last month that it was open-sourcing its video-generating AI models, while Tencent released five new open-source AI models in early March that can transform text and images into 3D visualizations.
Smaller AI firms are also following the trend. Chinese company Manus AI, which recently unveiled an AI agent that it claims is superior to OpenAI’s Deep Research, said it would go open source. Zhipu AI, one of China’s leading AI startups, announced on WeChat this month that 2025 would be the “year of open source.”
Ray Wang, chief analyst and founder of Constellation Research, told CNBC that the companies were forced to do this after DeepSeek released its AI models. “Because DeepSeek is free, no other Chinese competitors will be able to charge for the same thing. To compete, they will have to move to open-source business models,” Wang said.
Tim Wang, managing partner at technology hedge fund Monolith Management, called what’s happening in the AI market an “Android moment,” meaning that when Google made the source code for its operating system available to everyone, it fueled innovation and the growth of an ecosystem of apps other than Apple’s.