CHAMELEON EARLY FUSION MULTI-MODAL

Summary from the TLDR AI newsletter

Meta has announced a natively multi-modal model called Chameleon that operates on images and text simultaneously like GPT-4o. It outperforms many alternative models. The Meta team has since far surpassed this work on internal models.