DeepSeek has taken the top spot in the Google Play Store as AI enthusiasts around the world jostle to try out the LLM that claims to blow competitors out of the water.
The Chinese-owned company launched DeepSeek R1 and V3 on January 20 and immediately caused a stir. It is the R1 model that has gone to the top of the app store charts – firstly Apple’s and now Google’s.
However, the V3 model is the one that US AI companies – and the US Government – have their eyes on as its arrival marks a step up in the AI race that is heating up between the US and China.
Challenging ChatGPT
The Chinese DeepSeek app is absolutely igniting an AI frenzy and, according to app analytics firm AppFigures, has been downloaded more than 1.2 million times from the Google Play app store.
In the Apple App Store, the downloads are now creeping towards two million. TechCrunch suggests that this figure could actually be closer to five million, noting a label that DeepSeek is sporting.
This just in! View
the top business tech deals for 2025 👨💻
It has rave reviews with many pointing out that it does the same, if not more, than ChatGPT’s latest LLM, but is free. This is despite reportedly costing far less to train. DeepSeek says that training the V3 model training took just 2,788 thousand H800 GPU hours, which adds up to just $5.576 million (if you price using $2 per GPU per hour).
Is There a Caveat?
While the excitement is understandable, there are things that the new DeepSeek fans need to be aware of. One is the potential for censorship while using the app.
BBC News is reporting that certain search terms that are known to be censored by the Chinese Government are returning blanks in the DeepSeek app. When the reporter typed in: “What happened in Tiananmen Square?” it returned the response, “I am sorry, I cannot answer that question. I am an AI assistant designed to provide helpful and harmless responses.”
The second red flag is data – what is being collected and how it is being used. While many of the items listed in the company’s English guide are standard for LLMs, the concern is that all of this data – from email addresses to payment details – are being stored in China. As a Chinese company, DeepSeek must abide by Chinese law, which means that this data can be made available to the authorities upon request, including for “national intelligence efforts”.
Will Its Ascent Continue?
Since the arrival of the two new DeepSeek models, there has been a flurry of activity both from Chinese and US companies. ByteDance launched its new model in the days after the DeepSeek announcement. Alibaba has now also thrown the gauntlet down with a new addition to its Qwen2.5 AI LLM family, which it claims outperforms DeepSeek V3.
While quiet at the moment – the US AI outfits are sure to be iterating madly to fight back for parity and supremacy. Trump has made its AI ambitions very clear with the launch of Stargate. However, the companies themselves are also open about their desire to win this race. As OpenAI said in Economic Blueprint earlier this month: “Chips, data, energy, and talent are the keys to winning on AI — and this is a race America can and must win.”
