DeepSeek Launches Multimodal Vision Model V4-Flash-Vision-Exp, Opens API Services

On August 21, according to DeepSeek's official account, DeepSeek has launched its multimodal model V4-Flash-Vision-Exp and opened API services. Building on the capabilities of the original V4-Flash, this model introduces image understanding and vision generation features, supporting multimodal input and output scenarios such as text and images. It provides developers with API interfaces for inference and generation capabilities, marking the expansion of its model lineup from a text-only model to a multimodal product matrix.