Jin’s Daily AI Trivia: DeepSeek-R1-0528 Launched! 🚀

Jin’s Daily AI Trivia: DeepSeek-R1-0528 Launched! 🚀

The so-called “minor” upgrade of the DeepSeek-R1 model is now live — and it’s anything but minor!

Notably, DeepSeek-R1’s Chain-of-Thought (CoT) behavior appears to have improved significantly, showing more structured and coherent reasoning. Some early users have even observed deeper logic patterns, reminiscent of Google’s Gemini 2.5 Pro models, along with better writing quality and a more deliberate, thoughtful tone.

In my early testing, this “minor” upgrade performs on par with — or even better than — OpenAI’s o3-high.

The craziest part? It’s FREE and available now in the Hugging Face repo.