DeepSeek R1 Explained: China’s Ambitious Answer to ChatGPT, Gemini, and Why It’s in the News
A new big language model coming out of China has become the talk of the tech town. What is DeepSeek and is it better than ChatGPT and other AI models?
When OpenAI’s ChatGPT came out about two years ago, many thought that China was left behind when it came to AI. First, Microsoft and OpenAI and then companies like Google, Meta, Amazon ushered in the era of artificial intelligence. Chinese players like Alibaba, Baidu started investing billions of dollars but it seemed like they were lagging behind. But that’s not the case anymore, as a new big language model coming out of China has become the talk of the tech town.
What is DeepSeek and is it better than ChatGPT and other AI models? Here is a detailed description:
What is DeepSeek R1
DeepSeek R1, a cutting-edge reasoning model developed by Chinese AI startup DeepSeek, was launched earlier this month and has garnered significant attention for its exceptional performance and competitive pricing. DeepSeek R1 aims to enhance reasoning and analytical capabilities, setting it up as a formidable competitor to other leading AI models like OpenAI’s O1 and ChatGPT.
DeepSeek R1, an advanced language model, is built using a hybrid architecture similar to its predecessor, V3. Yes, it is not the Chinese startup’s first model but is better in almost every aspect.
It extensively incorporates reinforcement learning (RL) and chain-of-thought reasoning to enhance the accuracy of its responses. The model includes two versions: DeepSeek-R1 and DeepSeek-R1-Zero. Notably, the latter undergoes unsupervised fine-tuning, demonstrating remarkable reasoning capabilities.
What is the origin of DeepSeek?
Although R1 has garnered a lot of attention, DeepSeek itself remains a lesser-known entity. According to a report by MIT Technology Review, the company, headquartered in Hangzhou, China, was founded in July 2023 by Liang Wenfeng, a graduate of Zhejiang University specializing in information and electronic engineering. DeepSeek was incubated by High-Flyer, a hedge fund founded by Liang in 2015. Like OpenAI’s Sam Altman, Liang aims to develop artificial general intelligence (AGI) – an advanced AI capable of performing a wide range of tasks at or beyond human-level efficiency.
Why is DeepSeek’s popularity growing?
It’s always about the money, right? One of the notable features of DeepSeek R1 is its cost-effectiveness. Unlike OpenAI’s o1, which charges $15 per million input tokens and $60 per million output tokens, DeepSeek R1 offers significantly lower prices at $0.55 per million input tokens and $2.19 per million output tokens. This makes it an attractive option for developers, researchers, and organizations seeking cost-effective AI solutions.
What’s even more impressive is that, according to DeepSeek, it took the startup about two months to develop the model. While OpenAI, Google, Microsoft are spending billions of dollars in AI model development, DeepSeek invested ‘only’ $6 million to build its latest model.
In terms of performance, DeepSeek R1 has demonstrated results comparable to OpenAI’s o1 in various benchmarks, including math, coding, and logic tasks. Notably, it even outperforms OpenAI’s o1 in some areas, such as coding tasks, where it achieves a remarkable 97% success rate.
In addition, DeepSeek has introduced six compact versions of its R1 model, designed to run efficiently on laptops. The company claims that one of these smaller models outperformed OpenAI’s o1-mini in specific benchmarks. “DeepSeek has effectively replicated the o1-mini and made it open source,” Perplexity CEO Arvind Srinivas said in a post on X.
Microsoft CEO Satya Nadella commented in reference to DeepSeek – “I think we should take developments from China very seriously.”
Discussion about DeepSeek
The recent launch of DeepSeek R1 has sparked discussions on social media platforms, with many users sharing their experiences and comparative analysis between DeepSeek R1 and other AI models. Notably, AI and technology educator Paul Couvert outlined the seamless performance of DeepSeek R1 version 1.5b on his smartphone, which surpassed GPT-4O and Cloud 3.5 Sonnet in mathematical calculations. Furthermore, another X user – ZeroEdge – made a comparative evaluation of the models’ capabilities in executing a rotating triangle with a red ball, demonstrating the superior results of DeepSeek R1.
A more open(source) China?
Chinese companies are increasingly adopting open-source practices with a focus on efficiency. For example, Alibaba has introduced more than 100 open-source AI models in recent months, supporting 29 languages and addressing diverse applications such as coding and mathematics, according to an MIT Technology Review report.
A report published by China’s Academy of Information and Communication Technology, a state-affiliated research institution, highlights the global scale of AI development. As of last year, there were 1,328 large language models worldwide, of which 36% were from China. This solidifies China’s position as the second-largest player in AI development, trailing only the United States. Although it still lags behind the US in AI development, companies like DeepSeek could give China an edge and make the US worry about what might happen next. With its rapid growth, competitive pricing, and open-source initiatives, DeepSeek is set to make its presence felt in the global AI landscape and also signals China’s growing influence in the field.
How to access DeepSeek R1?
Users can visit the DeepSeek chat interface at chat.deepseek.com. A valid email address is required to sign up, and clicking the “DeepThink” option on the homepage grants access to the platform.
For developers wishing to integrate DeepSeek-R1 into their applications, API access is available through the DeepSeek developer portal. After obtaining an API key, developers can set up their environment using tools such as Python’s requests library or the OpenAI package. The API client should be configured with the base URL: api.deepseek.com.
