Chinese AI startup DeepSeek has quietly released an updated version of its R1 reasoning model, named R1-0528, on the developer platform Hugging Face, intensifying competition with U.S. tech giants like OpenAI and Google. Although the company has not yet issued an official public statement, the update is already making waves across the global artificial intelligence community.
DeepSeek R1-0528 Released on Hugging Face
The R1-0528 update went live in the early hours of Thursday, with no detailed documentation or benchmarks from DeepSeek itself. However, the LiveCodeBench leaderboard, a respected benchmarking tool developed by researchers from UC Berkeley, MIT, and Cornell, ranks R1-0528 just behind OpenAI’s o4 mini and o3 models, while outperforming Grok 3 mini and Alibaba’s Qwen 3 in code generation tasks.
While the company described the release internally as a “minor trial upgrade,” the model is now open for testing by developers, according to communications shared on WeChat.
Challenging Western Dominance in AI
This latest move by DeepSeek further fuels speculation that U.S. export controls are no longer a serious barrier to China’s rise in AI innovation. Earlier this year, DeepSeek disrupted global markets with the original R1 model, which delivered performance on par with leading U.S. models at significantly lower costs.
The January 2025 release of R1 rattled markets, sending tech shares outside China plummeting and challenging the conventional belief that scaling AI requires massive investment and computing power. Since then, major Chinese tech players like Alibaba and Tencent have rolled out their own models, claiming superiority over R1.
Live Code Bench Rankings & Global Impact
DeepSeek’s R1-0528 being ranked closely behind OpenAI’s o4 mini and o3 reinforces the startup’s position among the global AI elite. In a rapidly evolving AI race, even slight performance differences in code generation and reasoning models can make significant impacts on market adoption and developer interest.
Meanwhile, OpenAI has responded with price cuts and more accessible model versions, such as the o3 Mini, while Google’s Gemini has rolled out discounted access tiers to counter increasing competition from Chinese firms.
DeepSeek’s Roadmap: R2 on the Horizon
The AI community is now eagerly awaiting the launch of DeepSeek R2, expected to be the full successor to the original R1 series. Sources previously told Reuters that R2 was initially scheduled for a May release, though the company has remained tight-lipped about specific launch dates.
In March, DeepSeek also upgraded its V3 large language model (LLM), further diversifying its AI portfolio and offering enhanced capabilities across natural language processing and multi-modal tasks
DeepSeek’s quiet release of R1-0528 underscores a critical shift in the AI power dynamic between China and the United States. As the LiveCodeBench rankings place the updated model ahead of notable competitors like xAI and Alibaba, it signals that China’s AI ecosystem is maturing rapidly and doing so with leaner infrastructure and lower costs.
All eyes are now on the expected DeepSeek R2 launch, which could redefine the next phase of AI development and further challenge U.S. tech dominance.
