Categories
Chinese AI company DeepSeek has released an upgraded version of its most popular reason-based model, R1, with the codename R1-0528. The updated model went publicly accessible on May 28th, 2025, via Hugging Face. Hugging Face Platform, from which it is now accessible by researchers, developers, and AI enthusiasts across the world. While DeepSeek has yet to make an official announcement, the new model has already received a lot of interest from the worldwide AI community.
Initial tests indicate that R1-0528 significantly improves over the previous version. LiveCodeBench, developed with UC Berkeley, MIT, and Cornell, confirms this. The model matches OpenAI’s O4 Mini and O3 in code generation tasks. It also beats xAI’s Grok-3 mini and Alibaba’s Qwen-3 in many reasoning and coding tests. This makes R1-0528 one of the top open-source models available today.
While DeepSeek is describing the update as a trial update, it could have a giant impact. The original R1 model was launched at the beginning of January 2025 and completely transformed the AI landscape, offering high performance yet at one-third the price of models made in America. It debunked the notion that China’s AI progress would be hampered because of U.S. export controls on top-of-the-line technology. Yet, DeepSeek demonstrated that efficient models and smart engineering can bridge any gap, without having access to the most recent chips.
Users of the R1-0528 model have already seen clear improvements in response quality. They also report better reasoning accuracy and output formatting. This refinement strengthens DeepSeek’s reputation for building powerful, efficient models. Resource efficiency is now a key factor in large-scale AI deployment.
This release furthers the trend of openness that DeepSeek has been pursuing and reflects the practices of its international competitors like Meta as well as Mistral. It also demonstrates China’s increasing capabilities in the field of frontier AI advancement, suggesting that competition in this area will only increase.
As AI continues to transform industries and bring new opportunities to the market, DeepSeek’s rapid growth is a sign of an evolution. The leadership in AI is now multipolar. The R1-0528 public release confirms this trend. The model is now available to test using Hugging Face. This move is a step to strengthen DeepSeek’s position in the ever-changing AI landscape.
For the latest news, be sure to follow AtoZTechWorld or visit Facebook page.
