The power dynamics in the generative AI world are shifting dramatically with the rise of DeepSeek, a Chinese-developed AI assistant, which now claims the top spot on the U.S. iPhone charts.
Developed by a relatively unknown startup from Hangzhou, DeepSeek has managed to surpass well-established competitors, pushing Industry behemoth ChatGPT down a notch to number three.
No other AI chat app features in the top 100 at the time of writing.
DeepSeek’s ascent is powered by its DeepSeek-V3 model, boasting performance benchmarks that rival the most advanced closed-source AI models worldwide.
This achievement is remarkable in the context of Washington’s stringent export controls aimed at curbing China’s access to cutting-edge chips, including the top line from Nvidia, required to train complex AI systems.
DeepSeek’s developers claim to have trained their model efficiently, utilizing Nvidia’s H800 chips for less than $6 million in costs — a stark contrast to the multi-million dollar budgets often associated with AI training.
This shakeup in the app sphere interestingly parallels ongoing discussions in the U.S. political arena regarding another Chinese entity—TikTok.
President Donald Trump has asserted he’s engaged in conversations to potentially purchase the app. Speculations are abound that tech giant Oracle might play a critical role in these negotiations.
As TikTok potentially trades hands to placate national security concerns, DeepSeek is capturing U.S. user interest, drawing admiration from U.S. tech leaders.
DeepSeek is an open-source model but the company behind it has also made a native chatbot available on chat.deepseek.com and the mobile app to let users experience the powerful model.
It is this chatbot that also highlights the battle over information control with training generative AIs.
DeepSeek for example refuses to answer any questions related to controversies that would rattle the incumbent Chinese government, especially the issue of Taiwan and incidents like the Tiananmen Square, as tested by Dzambhala.
Yet this concern isn’t only with Chinese apps. OpenAI also drew flak earlier, especially from Elon Musk, over its restrictions over politcally-charged topics under former President Biden.
Musk, criticizing OpenAI CEO Sam Altman and the company’s flagship ChatGPT interface for going “woke,” went on to a launch a significantly less censored generative AI “Grok.”
It would be certainly interesting to see how the US v/s China aspect of the AI battle evolves but as Meta’s Chief AI Scientist Yann LeCun notes, we must not loose sight of the significance of an open-source model outperforming the benchmarks of closed models funded by deep pockets.
Facebook parent Meta has launched its own series of open-source models under the Llamma banner.
Do you want to plan your financial future? Fire Fast by Dzambhala is the right tool for you.
Join the vibrant privacy-ensured Dzambhala community on
Want to give feedback on this story? Write to us.