Chinese AI startup DeepSeek is expected to launch its next-generation AI model, V4, in mid-February, featuring advanced coding capabilities, according to The Information citing sources familiar with the matter. Internal testing reportedly suggests V4 may outperform competitors such as Anthropic’s Claude and OpenAI’s GPT series in programming tasks.
The V4 model is designed to handle and process extremely long coding prompts, potentially benefiting developers working on complex software projects.
Headquartered in Hangzhou, DeepSeek has become a notable player in China’s AI ecosystem, particularly in supporting domestic AI chip development. Its previous models, DeepSeek-V3 and DeepSeek-R1, received attention from global tech executives for their performance.
The startup previously built a low-cost rival to ChatGPT and has faced scrutiny in some countries over security and privacy concerns. Recently, DeepSeek also revealed an energy-efficient AI framework to improve scalability and efficiency under hardware constraints, highlighting China’s broader efforts to advance large-scale AI amid semiconductor restrictions.
DeepSeek did not immediately respond to requests for comment.
