DeepSeek Rolls Out Update to AI Model That Shook US Tech Stocks Earlier This Year

Chinese AI startup DeepSeek quietly released an update to its R1 reasoning model early Thursday, stepping up the competition with major US players like OpenAI. Though the company hasn’t made an official public announcement, the update—dubbed R1-0528—is now available on the developer platform Hugging Face for early testers.

While DeepSeek has yet to publish detailed info or comparisons about the new model, independent benchmarks are already making waves. The LiveCodeBench leaderboard—a respected benchmark created by researchers at UC Berkeley, MIT, and Cornell—places DeepSeek’s updated R1 just behind OpenAI’s powerful o4 mini and o3 models in code generation. Impressively, it ranks ahead of xAI’s Grok 3 mini and Alibaba’s Qwen 3, signaling DeepSeek’s rising influence in the AI race.

A Quiet But Powerful Move

Earlier reports from Bloomberg revealed that DeepSeek described this release as a “minor trial upgrade,” with users encouraged to begin testing immediately. This update follows a string of breakthroughs from DeepSeek that have already disrupted industry assumptions.

Back in January, DeepSeek stunned the tech world by launching the original R1 model, challenging the belief that China’s AI progress was being held back by US export controls. The model matched or outperformed some of the leading US AI models — all while running at a fraction of the cost.

That announcement sent tech stocks outside China tumbling, rattling markets and raising questions about the future landscape of AI development. Since then, Chinese giants like Alibaba and Tencent have raced to release models aiming to surpass DeepSeek’s achievements.

The Global AI Race Heats Up

The AI arms race is intensifying globally. Google has rolled out discounted access tiers for its Gemini model, while OpenAI responded by lowering prices and introducing an efficient o3 mini model designed to use less computing power.

Meanwhile, DeepSeek is rumored to be preparing a major release: R2, the highly anticipated successor to R1. According to Reuters sources from March, R2 was initially expected to debut in May. The company also pushed out an upgrade to its V3 large language model earlier this year.

What This Means for the Future

DeepSeek’s advancements signal a shift in the global AI landscape. Their ability to develop high-performing AI models at a much lower cost challenges long-standing assumptions about the scale of investment and computing power needed to compete with US tech giants.

As DeepSeek and Chinese tech giants continue to push boundaries, the AI competition is entering a new phase — one that could reshape innovation, pricing, and accessibility worldwide.