Pixtral Large, the newly released multi-modal open model, is gaining attention for its SOTA performance, comparable to GPT-4o.
A quick trial shows the model capable of Cantonese conversation and Chinese text OCR/understanding in images. It looks promising. #MultiModalModel
