DeepSeek launches AI engine with same features as ChatGPT: They are offering a $200 service for free

The news that DeepSeek has developed an AI capable of rivaling OpenAI broke following President Donald Trump's announcement last week that he would invest $500 billion in the AI industry. R1, which claims to provide superior performance in languages other than English, was developed for 5.5 million dollars.

DeepSeek AI models, which were disconnected from the internet in 2021, do not contain up-to-date data. However, in terms of performance, the DeepSeek AI engine is in the same class as ChatGPT o1.

Unlike OpenAI, DeepSeek does not pull information from the internet in real-time. The internet data used by the AI model does not cover the period after 2021. Nevertheless, DeepSeek has the same functions as ChatGPT o1 in areas that do not require working with current news and data.

Speaking to MIT Technology Review, Hancheng Cao from Emory University made the following statement regarding R1:

''This could be a great opportunity, especially for researchers and developers with limited resources in the Global South.''

US SANCTIONS LED TO THE DEVELOPMENT OF R1

DeepSeek's success is remarkable given the increasing US export restrictions on cutting-edge chips for Chinese AI companies. However, recent reports show that these restrictions are not working as intended.

According to MIT Tech Review, the sanctions, ''rather than weakening China's AI capabilities, are pushing companies like DeepSeek to innovate in ways that improve efficiency, resource pooling, and collaboration.''

The fact that the Chinese company had to reorganize its training process to create R1 is related to the sanctions imposed on China. Nvidia, the world's largest chip manufacturer, produces GPUs for the Chinese market that have half the performance of those it offers to the rest of the world.

The Chinese AI producer had to change its AI development process to reduce the load on the GPUs. Zihan Wang, a PhD student in computer science at Northwestern University, suggested that thanks to this change, DeepSeek was able to produce an AI equivalent to GPT o1.

OpenAI charges $200 per month for its new o1 model

THE 'SIMPLE ENGINEERING' BEHIND DEEPSEEK

DeepSeek R1 has been praised by researchers for its ability to handle complex reasoning tasks, particularly in mathematics and coding. The model uses a 'chain of thought' approach similar to that used by ChatGPT o1. This allows R1 to solve problems by processing queries step-by-step.

Dimitris Papailiopoulos, a lead researcher at Microsoft's AI Frontiers research lab, states that what 'surprised' him most about R1 was its 'simple engineering.' Papailiopoulos noted the following on the subject:

'Instead of explaining every logical step in detail, DeepSeek targets correct answers and significantly reduces computation time while maintaining a high level of efficiency.'

THEFT ACCUSATION AGAINST DEEPSEEK

DeepSeek also released six versions of R1 that are small enough to be run locally on laptops. It is claimed that one of these models outperforms OpenAI's o1-mini on certain benchmarks. Despite these developments, a theft accusation against DeepSeek has also come from the US.

Perplexity CEO Aravind Srinivas wrote in a post on his social media account, 'DeepSeek largely copied o1-mini and open-sourced it.'

Srinivas's 'theft' accusation was made before the US company had begun any publicly known investigation into the matter. Srinivas also did not respond to requests for comment from the US press.

WHAT IS DEEPSEEK?

Despite the debate surrounding R1, DeepSeek is relatively unknown. Based in Hangzhou, China, the company was founded in July 2023 by Liang Wenfeng, a Zhejiang University graduate with a background in information and electronic engineering, and is backed by 'High-Flyer,' a hedge fund Liang founded in 2015.

Similar to OpenAI CEO Sam Altman, Liang aims to create artificial general intelligence (AGI), a type of AI that can perform or exceed human capabilities across various tasks.

After a two-year research and development process, the Chinese AI company managed to produce a model equivalent to the AI of the US-based ChatGPT. Against its competitors who spend billions of dollars on AI research, DeepSeek achieved the same success with a budget of 5.5 million dollars.