Chinese AI startup DeepSeek has launched its latest large language model, DeepSeek V4, in a fresh challenge to major US players such as OpenAI, Google and Anthropic. The release strengthens China’s growing push to compete in the fast-moving artificial intelligence race.
The company introduced two versions of the model, V4 Pro and V4 Flash. While the Pro version is aimed at advanced reasoning, coding and research tasks, the Flash version is designed to be faster and cheaper for lighter everyday use. Both versions are currently available in preview mode.
DeepSeek said the new model can process very large amounts of text at once, allowing users to analyse long reports, documents and code files more efficiently. This could make the model attractive to businesses, researchers and developers handling large workloads.
The company also claimed improved performance in coding, logic and knowledge-based tasks compared with earlier versions. Independent testing of these claims is still awaited, but the launch has already drawn attention across the tech industry.
Another key feature is compatibility with Huawei’s Ascend chips, showing China’s efforts to reduce dependence on US chipmaker Nvidia amid export restrictions.
DeepSeek gained global visibility last year by offering lower-cost AI models that surprised the market with strong performance. Its latest launch is expected to increase pricing pressure on bigger rivals and expand AI choices for users worldwide.
The V4 release highlights how competition in artificial intelligence is no longer limited to Silicon Valley, with Chinese firms now emerging as serious challengers.
Also Read: Infosys gives ₹51.75 cr stock grant to CEO Salil Parekh