Google introduces Gemini 1.5 Flash, Gemma 2.0 and shares glimpse of Project Astra

Trained using a process termed “distillation”, Google says Gemini 1.5 Flash excels at “summarisation, chat applications, image and video captioning, data extraction from long documents and tables, and more.”

The tech giant also announced Gemma 2.0 and unveiled PaliGemma, the company’s first vision language model inspired by PaLI-3. As for Gemini 1.5 Pro, Google claims the latest version of the large language model can now follow more complex and nuanced instructions and understand image and audio for videos uploaded in Google AI Studio.

The Google DeepMind team also shared some information about Project Astra, a work-in-progress project that aims to build universal AI agents that excel in certain tasks. The company says these agents were trained on Gemini and can process information faster by encoding video frames and combining the video and speech input into a time of events. While the tech giant did not go into specifics, it looks like some of them will be coming to the Gemini app sometime later this year.

In its official release, Google stated that it has made significant strides with the Gemini models and is always seeking to push the boundaries of innovation. The company said that its dedication to continuous improvement allows it to explore new ideas and discover exciting new applications for the Gemini family.