What is Deepseek? Chinese AI startup that has taken over ChatGPT on App Store | Mint
Source: Live Mint
Chinese AI startup DeepSeek has challenged the dominance of top AI companies with its latest large language models, which offer similar performance to the latest offerings from Meta or OpenAI, but at a fraction of the cost. DeepSeek’s free app has also taken over ChatGPT on Apple’s App Store in several regions including the US, UK, Australia, Canada, China and Singapore.
What is DeepSeek?
DeepSeek was founded in 2023 by Liang Wenfeng, who is also the head of Chinese quant hedge fund High-Flyer. The company shot to fame last month after various benchmarks showed that its V3 large language model (LLM) outperformed those of many popular US tech giants, while being developed at a much lower cost.
DeepSeek’s R1 language model, which mimics aspects of human reasoning, also matched and outperformed OpenAI’s latest O1 model in various benchmarks.
At the time of writing, both DeepSeek R1 and DeepSeek V3 were ranked in the top 10 on Chatbot arena, a UC Berkeley-affiliated ranking that assesses the performance of leading chatbots. While Google’s Gemini 2.0 Flash and ChatGPT 4-o were both ahead of DeepSeek’s models, it ranked ahead of models from Elon Musk’s xAI and Amazon backed Anthropic.
While DeepSeek’s chatbots can compete with their Western counterparts on almost every metric, they are reluctant to answer questions that are sceptical of China. Any such question is met with a carefully crafted answer: “Sorry, that’s beyond my current scope. Let’s talk about something else.
Why is it a worry for US?
DeepSeek’s LLM’s are built on a much smaller costs and puts into question the ongoing belief that running AI models requires increasing amounts of computing power. In a paper last month, DeepSeek researchers stated that the V3 model leveraged the Nvidia H800 chips for training and incurred a cost of less than $6 million, a miserly sum compared to the billions that AI giants like Microsoft, Meta and OpenAI have committed to spend this year alone.
Moreover, the success of DeepSeek has also called into question the effectiveness of export curbs on advanced AI chips, a policy put in place by the Biden administration to slow down the growth of artificial intelligence in its adversorial countries.
DeepSeek’s recent advancement could lead to a decline in market share of top AI companies like OpenAI, Microsoft, Google and Meta. The effective pricing of DeepSeek could also lead to a reduction in the pricing of AI giants.
What is Silicon Valley saying about Deepseek R1?
Meta Chief AI scientist Yann LeCun wrote in a post on Threads, “DeepSeek has profited from open research and open source (e.g. PyTorch and Llama from Meta). They came up with new ideas and built them on top of other people’s work. Because their work is published and open source, everyone can profit from it. That is the power of open research and open source.”
Meanwhile, venture capitalist Marc Andreessen called DeepSeek a profound gift to the world. He wrote in a post on X, “Deepseek R1 is one of the most amazing and impressive breakthroughs I’ve ever seen — and as open source, a profound gift to the world. 🤖🫡”
Salesforce CEO Mark Benioff referred to the startup as ‘Deepgold’ in a post on X, writing, “Deepseek is now #1 on the AppStore, surpassing ChatGPT—no NVIDIA supercomputers or $100M needed. The real treasure of AI isn’t the UI or the model—they’ve become commodities. The true value lies in data and metadata, the oxygen fueling AI’s potential. The future’s fortune? It’s in our data. Deepgold. 😇”
