DeepSeek, a mobile AI application developed by a Chinese team, has overtaken OpenAI’s ChatGPT to claim the number one spot on the iPhone free app charts in both China and the United States. This achievement comes shortly after the launch of DeepSeek’s innovative “reasoning model,” DeepSeek R1, which allows the app to tackle complex logical and mathematical problems with ease. The breakthrough comes at a pivotal time for the AI industry, with DeepSeek positioning itself as a formidable competitor to established giants like OpenAI.
DeepSeek R1 Model: A Game-Changer
One of the key features of DeepSeek’s rise is its “DeepThink (R1)” mode, which displays the app’s “thinking process” before generating a response. This transparency allows users to follow the app’s reasoning, making it especially useful for intricate problem-solving. According to DeepSeek’s official website, the R1 model’s performance is comparable to OpenAI’s GPT-4, but at a fraction of the cost—approximately one-thirtieth of OpenAI’s operational expenses.
For the general public, the app and its web chatbot remain free to use, with only API calls for developers requiring payment. DeepSeek has also made its full-size model available for free download, enabling users with the necessary hardware to run it locally. For those with less powerful devices, scaled-down versions optimised for everything from ultra-thin laptops to high-performance gaming PCs are available.
Open-Source Approach and Industry Reactions
In addition to making its models freely accessible, DeepSeek has published a research paper detailing the development of the R1 model, allowing other developers to replicate the process with their own training data.
The release of the R1 model has sparked considerable attention within the AI community. Marc Andreessen, venture capitalist and co-founder of Netscape, praised the breakthrough as “one of the most amazing and impressive breakthroughs” he had ever seen, calling it “a profound gift to the world.”
Other figures in the AI space, including Perplexity CEO Aravind Srinivas and Nvidia’s Jim Fan, have lauded DeepSeek’s open-source approach, noting that it empowers the broader AI community and upholds the values of open research.
Meta’s Yann LeCun, however, cautioned that DeepSeek’s success should not be seen as evidence of China surpassing the U.S. in AI, but rather as an example of open-source models outpacing proprietary systems.
China’s AI Innovation: A Shift in Perspective
DeepSeek’s success also underscores a larger shift in China’s role in global AI development. In a July 2024 interview with 36kr.com, DeepSeek CEO Liang Wenfeng argued that China is moving beyond the stereotype of merely adapting U.S. innovations. As China’s economy grows, he said, the country is poised to become a global leader in innovation rather than relying on the work of others.
Liang emphasised that the development of DeepSeek’s earlier models, including V2, was entirely driven by local talent, without any overseas Chinese involvement. “Perhaps the top 50 talents in this field are not in China,” Liang stated. “But we can cultivate our own.”
As DeepSeek continues to break new ground in AI, it seems the company is on track to challenge the status quo, with China’s role in the global AI landscape becoming increasingly prominent.