DeepSeek Multimodal Team Researcher Teases New Model: 「Now, we see you」

According to Dongcha Beating monitoring, DeepSeek's multimodal team researcher Xiaokang Chen (@PKUCXK) posted on X, saying, "Now, we see you." The post included an image comparing two DeepSeek whale mascots: the whale on the left had its eyes covered, while the one on the right had its eyes open.

Considering his personal bio "Researcher in Multimodal Team @deepseek_ai" and the metaphor of "seeing," it seems to suggest that DeepSeek is hinting at an upcoming release of a new multimodal (vision) large model.

The DeepSeek V4 released on April 24 was a text-only model that did not support image inputs. If a multimodal model is launched, it will be a separate product line. DeepSeek has not made any official announcements yet.