DeepSeek added an image recognition mode to its chatbot and began offering it to select users on its website and mobile app for beta testing, according to a social media post by multimodal team leader Chen Xiaokang on Wednesday. [1]
The new mode appears alongside the chatbot’s “expert” and “flash” options. [1] The rollout comes just days after DeepSeek released its new flagship model V4 and after extensive price cuts. [1]
DeepSeek is based in Hangzhou. The company drew global attention in January 2025 for its model’s reasoning performance and low cost. [1]
The image recognition feature marks another step in DeepSeek’s push into multimodal tools after it introduced the “expert” and “flash” chat modes earlier this month. [1]
