Liang Wenfeng internally announced that DeepSeek V4 will be released in late April.

According to multiple informed sources revealed to the media, DeepSeek founder Liang Wenfeng recently disclosed in internal communication that the next-generation flagship model DeepSeekV4 will be officially released in late April. Recently, two interactive modes, "Quick Mode" and "Expert Mode," quietly went online on the DeepSeek website, supporting simple search and long-range, complex problem handling, sparking speculation in the industry about the release of the V4 model. Related news shows that DeepSeekV4 will not only bring trillion-parameter scale and a million-level context window but also achieve deep compatibility with domestic chips such as Huawei Ascend, marking a crucial step for the Chinese AI industry in the "de-CUDA" path. Foreign media reports claim that in preparation for the V4 release, domestic tech giants such as Alibaba, ByteDance, and Tencent have pre-ordered tens of thousands of new generation AI computing chips, planning to provide DeepSeek's new model through cloud services and integrate it into their own AI products. As a result, the prices of new AI chip products have recently risen by about 20%. As of the time of writing, DeepSeek has not responded to inquiries.