Chinese AI startup DeepSeek releases new model in rivalry development with US firms

Based in China Artificial intelligence (AI) Startup DeepSeek has released a new AI model in its pressure to compete with the most advanced US -based companies models.

The new Deepsekek model, DeepSeek-V3-0324, received a big update to its great language model V3, reinforcing performance in key aspects, including reasoning and coding, Reuters reported on Tuesday, citing benchmarks released on the AI ​​development platform.

The new model was available through the hug face platform.

DeepSeek has turned to head into silicone valley Because the company seems to fit the capabilities of chatbots like Chatgpt of Openai, but in some of the development costs.

What is China Ai startup DeepSeek?

Picture of DeepSeek

Chinese Ai DeepSeek is developing Chatbot to compete with Openai’s Chatgpt likes. (Getty Images, File / Getty Images)

The company said its V3 model was trained with computer power worth less than $ 6 million from the 2,000 NVIDIA H800 chips to achieve a level of performance at the most advanced models from Openai and META.

DeepSeek AI

The Chinese app for AI DeepSeek is seen at the Apple US App Store on January 27. (Christoph Dernbach / Alliance for Image through Getty Images / Getty Images)

The China -based firm The emergence raised questions about leading US tech companies investing billions of dollars in advanced chips and large data centers used to train AI models.

Get the fox business on the go by clicking here

It also serves as a “sputnik moment” for Ai race Between the United States and China after the perception that the United States had an advantage over its geopolitical rival in the field.

Eric Revel and Anders Hagstrom of Fox Business have contributed to this report.


Source link

Leave a Reply

Your email address will not be published. Required fields are marked *