Gemini and the multimodal turn

GGemini(Model), built by GGoogle(Company), is a family of multimodal models.

Unlike models that add vision as an afterthought, Gemini was pre-trained on text, images, audio, and video jointly.