Introduction
A look at the updated version of the AI model R1
DeepSeek has launched the updated version of the
artificial intelligence model R1 specialized in logical and deductive
thinking, that is, it is able to address complex tasks, which has
attracted attention globally and especially after outperforming the competing
models of major international companies specialized in artificial intelligence
such as OpenAI and Meta, as the launch took place without any official
announcement.
According to LiveCodeBench,
which tests AI models based on multiple criteria, the DeepSeek R1 model comes in close order to modern OpenAI
models such as O4-mini ando3.
“DeepSeek's recent upgrade is more efficient inference, stronger in math and
programming, and closer to leading models like o3 and Gemini,” Adina Yakivu, an
AI researcher at Hugging Face, told CNBC.
"The
model has seen significant improvements in the accuracy of the conclusion and
the reduction of the wrong information (hallucination), which shows that DeepSeek is no longer only
pursuing competitors but actually competing with them," the researcher
added.
Through what was the updated version exported?
I hope you have benefited from
this article. The article was written based on information from aitnews.
For more information, news and
technical topics, just follow our blog e-technook.com .
write a comment