DeepSeek’s V4 rollout arrived as the AI rivalry between China and the U.S. intensified, with the company presenting preview releases and positioning them as a step up in model capabilities and deployment flexibility. The update, announced Friday by the Chinese startup, targets users who have been comparing DeepSeek’s output with leading systems such as OpenAI’s ChatGPT, Anthropic’s Claude and Google’s Gemini.
In its preview release, DeepSeek said its V4 models are open-source and that they come in “pro” and “flash” versions. The company said the update brings improvements in knowledge, reasoning and in so-called “agentic” capabilities—meaning the model can carry out complex tasks and workflows autonomously.
DeepSeek’s V4 launch also underscored the timing pressures in the fast-moving market. The update had been keenly anticipated by users, and some industry analysts had expected it more than two months earlier at the start of the Lunar New Year, according to the Associated Press report.
On performance comparisons, DeepSeek said its “V4 Pro Max” version delivered “superior performance” on standard reasoning benchmarks relative to OpenAI’s GPT-5.2 model and Google’s Gemini 3.0-Pro. The company said its performance fell “marginally” short of GPT-5.4 and Gemini 3.1-Pro, and it described its release as arriving hours after OpenAI put out its new GPT-5.5 model on Thursday.
For “agentic” performance, DeepSeek said its “pro” version could outperform Claude’s Sonnet 4.5 and approach the level of Claude’s Opus 4.5, based on its own evaluation. DeepSeek said its “flash” version performs on par with the “pro” version on simple agent tasks and has reasoning abilities that “closely” approach the “pro” model.
The update also focused on technical capacity for running larger prompts. DeepSeek said both V4 “pro” and “flash” include a 1 million token context window, compared with the 128,000 token context window supported by its V3 model.
Chip compatibility was another part of DeepSeek’s positioning. Huawei said in a separate statement Friday that its Ascend chips and related technology are compatible with DeepSeek V4 models, and that the claim shows technical feasibility in operating outside the Nvidia-dominated computing ecosystem amid “sustained technological decoupling between China and the U.S.,” Marina Zhang said.
Analysts offered a range of assessments about how meaningful the V4 update will be for the competitive landscape. Lian Jye Su, chief analyst at the technology research and advisory group Omdia, said in the report that the benchmark results suggest DeepSeek V4 will be “very competitive” against its U.S. rivals. Marina Zhang described the V4 rollout as a “pivotal milestone for China’s AI industry,” particularly as global competition intensifies in efforts toward self-reliance in critical technologies.
Still, some observers cautioned against treating DeepSeek’s claims as a final verdict. Ivan Su, a senior equity analyst at Morningstar, said that while V4 is a “competent” follow-up, it is not as big a breakthrough as the earlier rollout of DeepSeek’s specialized “reasoning” model, R1. He also said DeepSeek’s own evaluation suggests capabilities “largely match” U.S. models on most fronts, but that independent evaluations are needed before final conclusions can be drawn.
The rollout came against a backdrop of dispute over how AI models are built and improved. In February, Anthropic accused DeepSeek and two other China-based AI laboratories of “industrial-scale campaigns” to “illicitly extract Claude’s capabilities,” using a technique called distillation, which Anthropic said involves training a less capable model on the outputs of a stronger one. OpenAI made similar allegations in a letter to U.S. lawmakers, and in recent remarks Michael Kratsios, a chief science and technology adviser to President Donald Trump, accused foreign tech companies “principally based in China” of distilling leading U.S. AI systems and “exploiting American expertise and innovation.” China’s embassy in Washington responded by describing the allegations as “unjustified suppression” of Chinese companies by the U.S., according to the AP report.