# Cross-modal Reasoning

Ristretto is an innovative vision-language model that employs dynamic image token deployment technology, allowing flexible adjustment of image token quantities based on task requirements, surpassing previous generations in performance and versatility.

ChatTime is a multimodal foundation model that unifies time series and text processing, featuring zero-shot forecasting capabilities and supporting dual-modal input/output for both time series and text.

ChemVLM is a multimodal large language model focused on applications in the chemical field, combining text and image processing capabilities.

Meta Chameleon is a hybrid-modality early-fusion foundational model developed by FAIR, supporting multimodal processing of images and text.

CogVLM is a powerful open-source vision-language model that achieves leading performance in multiple cross-modal benchmarks

Pix2Struct is an image encoder-text decoder model trained through multi-task learning for visual-language understanding tasks, specifically optimized for visual question answering on high-resolution infographics.