In a significant shift in the AI landscape, DeepSeek, a mobile AI application developed by a Chinese team, has surpassed OpenAI's ChatGPT to claim the top spot on the iPhone free app charts in both China and the United States. This achievement comes shortly after the release of its new \"reasoning model,\" DeepSeek R1.
Users activating the \"DeepThink (R1)\" mode are introduced to a display of the app's \"thinking process\" prior to generating responses. This feature enables DeepSeek to tackle complex logical and mathematical problems effectively. According to the company's official website, the R1 model's performance is \"on par with\" OpenAI-o1, while operating at approximately one-thirtieth of the cost.
Currently, DeepSeek offers its mobile app and web chatbot free for general use, with only API calls for programmers requiring payment. The company has also made its full-size model available for free download, allowing users with sufficient hardware to run it locally. For devices with less processing power, scaled-down versions of the model are offered, optimized for a range of hardware from ultra-thin laptops to high-performance gaming computers.
In addition to providing free access to its models, DeepSeek has published a research paper detailing the development of R1, enabling other developers to replicate the process using their own training data.
The AI Community Reacts
The release of the R1 model has stirred the AI industry, drawing reactions from prominent figures worldwide.
Marc Andreessen, venture capitalist and co-founder of Netscape, described DeepSeek R1 on social media platform X.com as \"one of the most amazing and impressive breakthroughs\" he has ever witnessed, calling it \"a profound gift to the world.\"
Aravind Srinivas, CEO of Perplexity, noted that \"DeepSeek has largely replicated OpenAI-o1-mini and has open-sourced it,\" highlighting the significance of the release.
Jim Fan, a senior research manager at Nvidia, commended DeepSeek for upholding the mission of conducting \"truly open, frontier research that empowers all.\"
Yann LeCun, Meta's chief AI scientist, commented that DeepSeek's success underscores how \"open-source models are surpassing proprietary ones,\" rather than indicating one country surpassing another in AI.
Kai-Fu Lee, former president of Google China, expressed on X.com that the DeepSeek release validates his belief in China's potential to excel in generative AI engineering.
DeepSeek CEO Highlights China's Innovative Potential
In a July 2024 interview with 36kr.com, DeepSeek CEO Liang Wenfeng emphasized that China must move beyond the stereotype of merely adopting foreign innovations. He stated that as the country's economy grows, it should transition into a global contributor to innovation rather than relying on advancements from elsewhere.
Liang highlighted that innovation arises from genuine curiosity, not just the pursuit of business success. He revealed that the DeepSeek team comprises young talent and that the development of the company's earlier V2 model did not involve any overseas Chinese contributors.
\"Perhaps the top 50 talents in this field are not in China,\" Liang said. \"But we can cultivate our own.\"
Reference(s):
cgtn.com
