InternVL 2

Open multimodal model rivaling GPT-4V — image, video, OCR and doc understanding.