DeepSeek rolls out V3 AI model updates in race against OpenAI. What's New?

DeepSeek, a Chinese artificial intelligence (AI) startup, has updated its V3 mode with the goal of improving programming capabilities. The V3-0324 update, which was first made public on Hugging Face this week, asserts that it will solve practical issues and establish standards for precision and effectiveness. In actuality, the V3 is an older DeepSeek platform. According to DeepSeek, it has significantly improved benchmark performance across a number of criteria.

Additionally, it promises improved multi-turn interactive rewriting, optimized translation quality and letter writing, improved report analysis requests with more detailed outputs, improved style and content quality when it comes to Chinese writing proficiency, and improved function calling accuracy, resolving issues from earlier V3 versions.

The discussion about whether cutting-edge platforms can be developed for significantly less money than the billions that US companies are spending on data center building has been sparked by the startup's AI services.