ClovisOCR

Deepseek v1Document analysis

A vision and OCR model that can extract text, convert documents to Markdown, locate elements in an image, and describe images visually. Supports DeepSearch modes with grounding.

Capabilities

InputWhat you can send

ContentPrompt + one or more images

Accepted formatsPNGJPGWEBPURLBase64

OutputWhat the model produces

Output formatsPlain textMarkdownCoordinates

ModesOCRDescriptionLocalization

Visual groundingYes

Custom promptYes

Compatible endpoints

POSTChat Completion (Vision)→

Documentation

OCR overview→