Amidst the existing powerful artificial intelligence models by big-tech companies of US and other western countries, a new AI model by a Chinese AI research lab has created waves in the Silicon Valley. Called “DeepSeek”, the Chinese AI model is on par to OpenAI’s state-of-the-art model for reasoning and the most differentiating part is that it’s ‘Free’ to use.
The capabilities and efficiency of DeepSeek made the companies and researchers in US’ to rethink what they’ve been doing so far in multiple aspects including the cost. With no preluding further, here are 7 important things that you may need to know about China’s “DeepSeek” AI.
7 Things to know about DeepSeek AI
#1 DeepSeek is an AI research lab based in Hangzhou, China. It was founded in 2023 by Liang Wenfeng, an engineer and entrepreneur with a background in AI and quantitative finance. Before creating DeepSeek, Liang led a hedge fund that used AI to analyze financial data. His team at DeepSeek consists mostly of fresh graduates from top Chinese universities.
#2 DeepSeek V3 and R1: DeepSeek’s models include the open-source DeepSeek-V3, which is the foundation for their other models. The company’s latest model, DeepSeek R1, is designed to mimic human thinking. DeepSeek claims that the R1 model offers performance on par with OpenAI’s latest technology in tasks such as math, coding, and natural language reasoning. It is said to provide its reasoning before delivering a response. According to the company, DeepSeek R1 bested other large language models in benchmarks, and is particularly skilled at mathematical, coding, and reasoning tasks. The DeepSeek-V3 model reportedly uses cheaper Nvidia H800 chips for training.
#3 DeepSeek’s AI assistant has been said to outperform models like ChatGPT, Gemini, and Claude in terms of performance. It is a free app that has topped the charts on Apple’s App Store in the US, surpassing even ChatGPT in downloads. In tests, DeepSeek claimed that its model outperformed OpenAI’s latest ChatGPT o1 model on key benchmarks.
We’ve also made a Collection of Top-10 Google’s AI tools for you. Check it.
#4 The DeepSeek-V3 model was reportedly developed for around $6 million, which is significantly less than the billions spent by other companies on similar models. The company uses existing technology and open-source code, making it more affordable to develop. DeepSeek’s approach has demonstrated that extremely clever AI does not have to be extremely expensive to train or use. The model’s API is also available for an extremely low cost — just $0.14 for one million input tokens, compared to OpenAI’s $7.5 rate for o1.
#5 DeepSeek models are open-source, with an MIT license, which means anyone can deploy, retrain, and customize the AI model for free. This approach is lowering the barrier to entry for AI innovation, allowing startups and smaller businesses to compete. The chatbot is available free of charge via app or web interface. DeepSeek is celebrated for its transparency, efficiency, and ability to make AI accessible to everyone.
#6 Disruption to the AI market & how US tech giants are feeling pressure: DeepSeek’s emergence has caused significant disruptions in the AI market, raising questions about the need for huge investments in high-end AI accelerators. It has challenged the belief that the US holds a dominant position in the AI space. The rise of this low-cost alternative has forced companies to reconsider their AI strategies. This has led to a decline in the stock prices of major US AI-related companies, including Nvidia, Meta, and Microsoft. In fact, Nvidia saw a record $465 billion drop in its market value due to investor concerns about DeepSeek.
#7 DeepSeek’s development comes as the US restricts the sale of advanced chip technology to China, but DeepSeek has managed to work around those restrictions by developing models that require fewer resources. Some experts believe that Liang Wenfeng built up a store of Nvidia A100 chips before they were banned from export to China. The success of DeepSeek could lead to a future where AI is more affordable and accessible.
How and Where to use DeepSeek AI?
DeepSeek AI can be accessed and used in the following ways, according to the sources:
- DeepSeek offers a ChatGPT-style chatbot that can be accessed through an app or web interface. This allows users to interact with the AI model for various tasks, similar to how they would use ChatGPT or Gemini. The chatbot is available for free.
- DeepSeek has a mobile app that is available for download on the Apple App Store. The DeepSeek app has become the top-rated free application in the US, surpassing ChatGPT in downloads.
- The DeepSeek chatbot is also accessible via a web interface, providing an alternative way to use the AI model.
- DeepSeek’s models are available via an API (Application Programming Interface), allowing developers to integrate the AI’s capabilities into their own applications and services. The API is available at an extremely low cost, with rates of just $0.14 for one million input tokens.
To begin using the DeepSeek AI, you will need to create an account or log in with your E-mail. For developers, the open-source framework with an MIT license, allows anyone to deploy, retrain, and customize its AI model for free. The company’s website provides access to its models as well as documentation for API use.
FAQ
What is DeepSeek?
DeepSeek is a Chinese AI research lab that has developed high-performing AI models at a fraction of the cost of its rivals.
Who founded DeepSeek and when?
DeepSeek was founded in 2023 by Liang Wenfeng, an engineer with a background in AI and finance.
What are DeepSeek’s key AI models?
DeepSeek’s key models include the open-source DeepSeek-V3 and the advanced DeepSeek R1 model, which is designed to mimic human thinking.
How does DeepSeek’s performance compare to other AI models?
DeepSeek’s AI assistant is said to outperform models like ChatGPT, Gemini, and Claude in various tasks, and the company claims R1 is on par with OpenAI’s latest technology.
How much did it cost to build DeepSeek AI?
The company’s researchers claim that their open-source DeepSeek-V3 model was developed for around $6 million. This is significantly less than the billions spent by rival companies. Some experts have estimated that OpenAI’s GPT-4 model cost more than $100 million to train3.
Is DeepSeek open-source and accessible?
Yes, DeepSeek models are open-source with an MIT license, making it free for anyone to use and customize, and the chatbot is available for free via app or web.
(For more such interesting informational, technology and innovation stuffs, keep reading The Inner Detail).
Kindly add ‘The Inner Detail’ to your Google News Feed by following us!
