Gemini: A Family of Highly Capable Multimodal Models
Gemini Team, Google DeepMind (incl. J. Adler) · arXiv preprint (2023)
The first Gemini models — a family of natively multimodal models with strong capabilities across image, audio, video and text understanding.