Chinese AI company DeepSeek on Friday released a preview of its much-anticipated next-generation AI model v4, saying the open-source model can compete with major closed-source systems from US rivals including Anthropic, Google and OpenAI.
DeepSeek says v4 represents a major improvement over previous models, especially in coding, a capability that has become central to AI agents and helped drive the success of tools like ChatGPT Codex and Cloud Code. The release is also a milestone for China’s chip industry, with DeepSeq clearly highlighting compatibility with domestic Huawei technology.
The release comes a year after DeepSeek shook up the US AI industry with R1, a model it claims was trained at a fraction of the cost of major US systems. DeepSeek has not disclosed the training cost of V4 or what hardware it was trained on. US authorities have accused the company of using banned Nvidia chips and Anthropic claims DeepSeek abused the cloud to improve its products.
<a href