According to market research, China's open-source AI models are driving the adoption and innovation of Chinese artificial intelligence, and industry insiders also call China's open-source AI models the "Android moment" in the field.
The open-source AI trend began with AI startup DeepSeek. Earlier this year, DeepSeek's R1 model shook the industry by challenging the US tech giant's dominance in AI with outstanding performance and lower cost. The significance of this model is far more than a breakthrough at the technical level, but also its successful practice of the open-source model. DeepSeek's R1 model is released under the "MIT License", an extremely permissive open-source license that allows developers to freely use, modify, and distribute, including for commercial purposes, which greatly stimulates innovation. DeepSeek also held an "Open-Source Week" event to expose more technical details of the R1 model, providing developers with an opportunity to learn more about it.
Driven by DeepSeek, many AI companies in China have embraced the open-source model. Tech giant Baidu is a prime example of this. In the past, Baidu has adhered to a proprietary licensing business model and has been cautious about open source. However, with the rise of open-source forces such as DeepSeek, Baidu has also made strategic adjustments. On March 16, Baidu released the latest AI model, Ernie 4.5, and inference model, Ernie X1, which are available to individual users for free. In addition, Baidu plans to open-source the Ernie 4.5 model series from the end of June, a shift that marks a major shift in the way China's AI industry develops, from closed to open, from focusing on proprietary technology to sharing innovations.
Figure: China's open-source AI model leads the trend (Source: Getty Images)
In addition to Baidu, tech giants such as Alibaba and Tencent are also actively following suit. Alibaba Cloud has open-sourced video generation AI models, while Tencent has released five open-source models that can convert text and images into 3D visuals. And smaller businesses like ManusAI are also adding to the open-source trend. ManusAI's co-founder of ManusAI's plan to open-source some of its models says it's all thanks to the support of the open-source community, and the company wants to give back to the community. Zhipu AI has declared 2025 the "Year of Open Source", demonstrating its strong belief in the open-source model.
The development of open-source AI in China has had a profound impact on the global AI landscape. Compared with the closed-source model of the United States, China's open-source route has unique advantages. Under the open-source model, developers can carry out secondary development and innovation based on existing models, accelerating technology iteration, lowering the threshold for innovation, and enabling more enterprises and individuals to participate in AI innovation. This model has driven the widespread adoption of AI technology, making AI no longer the exclusive domain of a few tech giants, but a tool for the masses.
Take DeepSeek's R1 model, for example, which is free on its own, but is monetized by charging relatively low API fees. This business model provides a new way of thinking for other companies to achieve sustainable business development while ensuring technological innovation. In contrast, U.S. companies such as OpenAI rely on closed-source code and high fees, facing high costs and limited adoption.
However, China's open-source AI development is not intended to compete with the United States in a simple way. As Alibaba Group Chairman Joe Tsai said, the significance of the open-source model is to empower everyone to use AI, driving more development, innovation, and popularization of AI applications. In fact, some companies in the U.S. have also benefited from China's open-source efforts, demonstrating that open-source AI is the result of global collaborative innovation, not a tool for competition.
