Gemini

Gemini is Google's family of natively multimodal AI models. Unlike legacy models that are trained solely on text and then "taught" to see images, Gemini was trained from the start on a diverse mix of text, images, audio, video, and code.

This native understanding allows it to seamlessly reason across different types of information—for example, watching a video and answering questions about the audio track—making it a uniquely powerful engine for complex, media-rich applications.