DeepSeek launches experimental multimodal model V4-Flash-Vision-Exp

DeepSeek has launched its experimental visual understanding model, DeepSeek-V4-Flash-Vision-Exp, on its API platform, supporting JPEG, PNG, GIF, and WebP formats.

The model can receive images via Base64, external URL, or Files API, with each image billed for a maximum of 384 tokens; vendor benchmark results still need to be distinguished from independent evaluations.

refer to:DeepSeek API Docs