DeepSeek launches DeepSeek-V4-Flash-Vision-Exp and Files API

The multimodal model is available via the API with images tokenized and billed at up to 384 tokens each under V4-Flash pricing.

Images can be supplied using base64, external URLs, or the new free Files API.
The Files API allows users to upload an image once and reuse it across multiple requests using a file_id.
DeepSeek Harness 0.1.1 includes out-of-the-box support for the model.
The model supports Chat Completions, Messages, and Responses endpoints with mixed text and image inputs.

Developers using DeepSeek's API can now run visual agent workflows and reduce bandwidth overhead by referencing uploaded image files directly.

Sources