ColPali

Vision-language retrieval model that indexes document pages as images — no OCR needed.