Chinese Start-Up DeepSeek Revolutionizes AI Model Training with New LLM - PRESS AI WORLD
PRESSAI
Recent Posts
side-post-image
side-post-image
Technology

Chinese Start-Up DeepSeek Revolutionizes AI Model Training with New LLM

share-iconThursday, January 02 comment-icon4 weeks ago 9 views
News sources:
scmpSCMP
Chinese Start-Up DeepSeek Revolutionizes AI Model Training with New LLM

Credited from: SCMP

  • DeepSeek is gaining recognition as "the biggest dark horse" in the open-source large language model (LLM) sector as of 2025.
  • The recently released DeepSeek V3 boasts 671 billion parameters and was developed at a fraction of the cost and time compared to major tech firms.
  • Despite US sanctions on advanced semiconductors, DeepSeek leveraged innovative architecture for efficient training using only 2.78 million GPU hours.
  • The start-up faced challenges, including instances of identity confusion with its AI model, which misidentified itself as ChatGPT.
  • DeepSeek’s advancements highlight the rising capabilities of Chinese AI firms in the competitive landscape.

Read more about DeepSeek here.

SHARE THIS ARTICLE:

nav-post-picture
nav-post-picture