DeepSeek Releases Upgraded AI Model, Narrowing U.S.-China Tech Gap
The Chinese startup's V3-0324 model enhances reasoning and coding capabilities under a permissive open-source license, intensifying global competition.
- DeepSeek's V3-0324 model, launched under an MIT open-source license, features improved reasoning, coding, and mathematical problem-solving abilities.
- The model's performance on key benchmarks, including a 50% improvement on the AIME test, underscores its technical advancements.
- DeepSeek's open-source approach has spurred a shift in China's AI industry, with major players like Baidu and Alibaba adopting similar strategies.
- Privacy and security concerns surrounding DeepSeek's models have led to investigations and potential restrictions in the U.S. and Europe.
- Apple CEO Tim Cook praised DeepSeek's models during a visit to China, fueling speculation about potential future collaboration.
PYMNTS 
BGR 
Forbes 
Benzinga 
CNBC 
South China Morning Post 
VentureBeat 
Gadgets 360 
9to5Mac 
The Independent 
Fox Business 
TechRadar 
MacRumors 
ZDNET 
