DeepSeek has claimed the top spot on the US App Store’s “Top Free Apps” chart, surpassing OpenAI’s ChatGPT. The Chinese company launched its open-source DeepSeek-R1 artificial intelligence (AI) model last week, which has demonstrated superior performance compared to OpenAI’s o1 AI model across multiple benchmarks. This development has ignited discussions in the tech community, particularly among Silicon Valley leaders, regarding the app’s rapid ascent. Importantly, the DeepSeek app remains entirely free for users, with no subscription options announced to date.
DeepSeek Overtakes ChatGPT
Since OpenAI introduced its ChatGPT iOS app in May 2023, it consistently ranked among the top free applications on the App Store and was previously recognized as the leading AI app on the platform. However, recently, the DeepSeek app has ascended to the number one position in the charts, a change credited to the debut of the DeepSeek-R1 AI model.
Details about the Hangzhou-based AI firm, founded in 2023, remain somewhat scarce despite its release of various open-source large language models (LLMs). While other US tech companies, like Meta, have ventured into open-source AI, none appear to match the capacity and functionality offered by DeepSeek’s models. The firm claims it developed the R1 model at a cost of $6 million (approximately Rs. 51.8 crore), a notable bargain for models of this magnitude.
Several notable figures in Silicon Valley have expressed their astonishment at the swift emergence of DeepSeek’s technology. Venture capitalist Marc Andreessen described the R1 AI model as “one of the most amazing and impressive breakthroughs I’ve ever seen.” Additionally, Aravind Srinivas, Co-Founder and CEO of Perplexity AI, extended his congratulations to the Chinese firm for achieving the milestone of being the first AI app to surpass ChatGPT.
Yann LeCun, a prominent AI figure and Chief AI Scientist at Meta, shared insights on LinkedIn, asserting that “open source models are surpassing proprietary ones.” Meanwhile, Holger Zschäpitz, a senior financial reporter for the German news channel Welt, voiced concerns that DeepSeek “could represent a significant threat to US equity markets.”
DeepSeek Is Not the Only One
While DeepSeek is breaking the trend of companies opting to open-source smaller, less capable AI models while keeping more advanced models proprietary, it is not the sole player in this space. Another Chinese company, Kimi AI, has unveiled its Kimi k1.5 AI model, claiming it is an “o1-level multimodal LLM that outperforms GPT-4o and Claude Sonnet 3.5 on various benchmarks.”
Significantly, Kimi AI has made its web-based chatbot freely accessible for unlimited use, mirroring DeepSeek’s approach. The model boasts capabilities such as real-time web searches, analysis of up to 50 files in a broad range of formats, and image understanding functionality. However, while its technical report is available on GitHub, the model itself is not currently released as open-source.
