A new super-powered, open-source AI model called DeepSeek R1 is rattling the industry this week, after it was unexpectedly dropped into the laps of artificial intelligence experts — and the world — with seemingly valid challenges to OpenAI's expensive AI model.
Released by Chinese AI startup DeepSeek, the DeepSeek R1 advanced reasoning model purports to outperform the most popular large language models (LLMs), including OpenAI's o1. According to the company, DeepSeek R1 bested these black box offerings in several important benchmarks, and has a particular talent at mathematical, coding, and reasoning tasks, Mashable's Stan Schroeder reports. Schroeder's own tests have shown that it holds its own against rival ChatGPT in complex coding tasks.
"In the background of all this are training costs which are orders of magnitude lower than for some competing models, as well as chips which aren't as powerful as the chips that are on disposal for U.S. AI companies," wrote Schroeder. "DeepSeek thus shows that extremely clever AI with reasoning ability doesn't have to be extremely expensive to train — or to use."
DeepSeek is free to use – how to try it now
For curious minds and those looking for open source alternatives to the industry's current major players: DeepSeek's chatbot offering is free to use on the web and now available for download on the Apple App Store.
In order to use the new, "highly-intelligent" DeepSeek-V3 model, users will need to create an account or log-in with their Google information.
The model's API is also available for extremely low cost — just $0.14 for one million input tokens, compared to OpenAI's $7.5 rate for o1. In addition to the full size (32-billion-parameter) RI model, DeepSeek offers smaller, distilled models ranging from 1.5 billion parameters to 70 billion, reports the Register. Create an account on DeepSeek's API platform to get started.
