Gemini is designed as a multimodal AI model family, capable of processing not just text but also audio, images, and videos, unlike its predecessor LaMDA.
Google's Gemini comes in various forms such as Ultra, Pro, Flash, and Nano, each serving distinct purposes from general use to more lightweight offline applications.
The ethics of using public data for training models like Gemini remain complex, as Google’s policy offers some legal protection to its cloud customers, but with significant limitations.
Gemini apps serve as interfaces connecting users to Gemini models, functioning similarly to platforms like ChatGPT, while maintaining a distinction between the models and their applications.
[
Collection
]
[
|
...
]