A young game developer, Annie is designing a new role-playing game comprising complicated narratives. She sought the help of a large learning language model (LLM) such as ChatGPT to generate an effective and interesting dialogue for the characters in the game.
However, the developer can’t afford to customise or experiment more dynamically using ChatGPT.
Finding herself at a crossroads, Annie then switched LLM choice to DeepSeek, a more affordable Chinese AI model which allowed her to tailor her requirements more flexibly.
Topping the App Store Charts on iPhones today is the Chinese rival of ChatGPT called DeepSeek.
Hailed as the “Temu of ChatGPT” by Financial Times, DeepSeek is an open-source large language model (LLM) that rivals ChatGPT-owner “OpenAI's Model o1.”
BBC reported earlier today that the Chinese AI app DeepSeek has overtaken ChatGPT and other rivals to become the top-rated free application on Apple's App Store in the US, UK and China.
According to DeepSeek’s researchers, its open-source model DeepSeek-V3 was developed for less than $6 million which is conspicuously less than the billions of dollars spent by the LLM’s rivals.
In an update, DeepSeek allegedly suffers a cyber attack. Reuters reported today that DeepSeek has temporarily limited registration due to a large-scale cyber attack on its services.
DeepSeek launched its latest model, DeepSeek-V3 on January 10, 2025.
Fiona Cincotta, senior market analyst at City Index said that the idea of a low-cost Chinese version hasn't necessarily been forefront, so it's taken the market a little bit by surprise.
"So if you suddenly get this low-cost AI model, then that's going to raise concerns over the profits of rivals, particularly given the amount that they've already invested in more expensive AI infrastructure."
Vey-Sern Ling, a Singapore-based technology equity advisor told the BBC it could "potentially derail the investment case for the entire AI supply chain".
What is DeepSeek?
DeepSeek is a China-based AI research lab that develops large language models (LLMs) such as its latest model Deep-Seek V3.
DeepSeek has often been pitted against OpenAI’s LLMs. In fact, the Chinese company’s website states:
“DeepSeek-R1 is now live and open source, rivalling OpenAI's Model o1. Available on web, app, and API.”
Since the LLM developing platform is open-source, it allows developers and researchers to freely access, customise, and deploy DeepSeek’s models into their software. The open-source nature allows for easy integration into different software and platforms, without being heavily restricted by Application Programming Interface (API) limitations or usage fees.
DeepSeek is available on the application stores on both Android and Apple devices, as well as on the web and API.
The company’s new AI chatbot, DeepSeek R1 seems to have also triggered market turmoil according to the Telegraph which could “wipe more than $1 trillion (£799bn) off technology companies at the opening bell in New York later.”
“The Nasdaq in New York plunged more than 3pc in premarket trading as DeepSeek said it had spent only $5.6m (£4.5m) developing its model - peanuts when compared with the billions US tech giants have poured into AI,” Telegraph reported.
Who Owns DeepSeek?
Liang Wenfeng, a Chinese AI entrepreneur owns DeepSeek. He is also the head of High-Flyer, the Chinese quant hedge fund.
The 40-year-old started DeepSeek in May 2023 in Hangzhou, a city in southeastern China. His hedge fund backed DeepSeek according to BBC.
A July 2024 interview with DeepSeek founder reported on The Chinese Academy following the launch of its open-source V2 model that this has “triggered a fierce price war in the large model industry.”
Wenfeng says that grabbing users wasn’t their primary goal.
“We reduced prices because, first, while exploring next-generation model structures, our costs decreased; second, we believe that both AI and API services should be affordable and accessible to everyone.”
Zihan Wang, a former DeepSeek employee and current PhD student in computer science at Northwestern University alluding to the development of R1, told MIT Technology Review that DeepSeek had to rework its training process to reduce the strain on its GPUs, a variety released by Nvidia for the Chinese market that have their performance capped at half the speed of its top products.
