The deepseek-v4-flash-vision-exp model processes images in JPEG, PNG, GIF, and WebP formats alongside text for tasks like image description, chart analysis, and text reading from screenshots. Images can be provided through base64 encoding, external URLs, or the Files API, with various size and format limits applying to each method.