Cephalo is a series of multimodal vision-language models designed to integrate visual and textual data for materials science applications. This addresses a significant challenge in the field and has the potential to improve research and applications in materials science.