### Check for existing issues

- [x] I have searched the existing issues and checked that my issue is not a duplicate.

### The Feature

Support OCR endpoints and Mistral through the Batches API. Mistral offers an OCR batch API (at a 50%-discounted rate) but it can’t be used through LiteLLM today — the Batches API admits only a fixed set of endpoints (chat/embeddings/completions) and providers. This feature consists of:

1. Enable Mistral in the Files API 2. Enable OCR in the Batches API, including a per-page cost path (OCR is billed per page, not per token) 3. Enable Mistral as a Batches provider

This follows the existing Bedrock batches pattern — no schema migration or new proxy endpoints needed.

### Motivation, pitch

Running Mistral OCR batches requires calling Mistral directly, bypassing LiteLLM's unified API, access control and cost tracking. Routing them through /v1/batches fixes that and unlocks Mistral’s 50%-discounted rate. It also makes OCR a first-class batch endpoint with per-page costing, so any future OCR provider (e.g. a vLLM-served OCR model) can plug into the same path.

### What part of LiteLLM is this about?

Proxy

### LiteLLM is hiring a founding backend e…