Chinese LLM DeepSeek-R1 threatens to disrupt US AI dominance with open-source solution

A Chinese company built an LLM solution called DeepSeek-R1 just five days ago, and the tremors are felt in the US, which is the undisputed leader in this space. The ChatGPT-like LLM, which is much cheaper and more efficient, is seen as something that can turn the tables against the resource-heavy model and provide an alternative and workable model for countries like India.

Even as the Big Tech (Google, Microsoft, Facebook, and Amazon) is involved in a fierce LLM battle, pumping billions of dollars into building improved versions of their solutions, DeepSeek promises to change the goalpost by building a better platform that can drastically reduce the cost of building LLMs.

While the US model comes with huge entry barriers for the Global South to develop their own ChatGPT-like solutions as they require huge computing, energy, human resources, and financial resources, the Chinese model, which is built on open-source software, seems to be replicable as it reportedly requires just 10 per cent of the resources.

Experts say the innovative reinforcement learning (RL)-based architecture of this open-source model has significantly advanced AI reasoning capabilities. 

The impact is so evident that Sam Altman, the Founder and Chief Executive Officer of OpenAI (which developed ChatGPT), announced the upgradation of the free version to its premium product. “The free version of the LLM is going to the tier of ChatGPT is going to get o3-mini,” he said in a post on ‘X’.

To make a distinction between the free and premium users, he announced 100 queries per day for its paid customers. He also said that the Operator tool (an AI agent that can work for you independently) will be available for the paid users.

“AI technology itself is powerful but adoption is limited by high costs, particularly in the Global South. DeepSeek has demonstrated a steep drop in training costs for frontier AI models. As DeepSeek’s models are available as open source, organisations can now consider AI use cases that were previously too expensive,” Kashyap Kompella, AI Industry Analyst, told businessline.

Depending on their use cases, customers can now choose between several viable open-source AI models – DeepSeek, Llama, Qwen, Mistral, and others. 

He, however, cautioned that the parent company of DeepSeek has deep pockets, a deep talent pool, and government support. “Even if the compute cost of training models may have come down, companies in the Global South will find it difficult to create cutting-edge models without all the ingredients in place,” he said. 

How the world reacted

DeepSeek has wowed the Internet. Social media platforms are abuzz with the news, with those who are following the LLM space closely talking about the new kid on the block. While geeks around the world took to platforms like ‘X’ and Reddit to discuss its pros and cons, others have started doing ‘How-To’ videos and tutorials, explaining people how to download it on their laptops and mobile phones to tap the power of the open source architecture.

_{Images are for reference only.Images and contents gathered automatic from google or 3rd party sources.All rights on the images and contents are with their legal original owners.}