got to check out Astra in a closed-door live demo last week. It was a shocking experience, but there’s a gulf between polished promo and live demo.
Astra uses Gemini 2.0’s built-in agent framework to reply questions and perform tasks via text, speech, image, and video, calling up existing Google apps like Search, Maps, and Lens when it must. “It’s merging together a number of the strongest information retrieval systems of our time,” says Bibo Xu, product manager for Astra.
Gemini 2.0 and Astra are joined by Mariner, a brand new agent built on top of Gemini that may browse the net for you; Jules, a brand new Gemini-powered coding assistant; and Gemini for Games, an experimental assistant which you could chat to and ask for suggestions as you play video games.
(And let’s not forget that within the last week Google DeepMind also announced Veo, a brand new video generation model; Imagen 3, a new edition of its image generation model; and Willow, a brand new sort of chip for quantum computers. Whew. Meanwhile, CEO Demis Hassabis was in Sweden yesterday receiving his Nobel Prize.)
Google DeepMind claims that Gemini 2.0 is twice as fast because the previous version, Gemini 1.5, and outperforms it on a variety of standard benchmarks, including MMLU-Pro, a big set of multiple-choice questions designed to check the talents of huge language models across a variety of subjects, from math and physics to health, psychology, and philosophy.
However the margins between top-end models like Gemini 2.0 and people from rival labs like OpenAI and Anthropic are actually slim. Today, advances in large language models are less about how good they’re and more about what you possibly can do with them.
And that’s where agents are available in.
Hands on with Project Astra
Last week I used to be taken through an unmarked door on an upper floor of a constructing in London’s King’s Cross district right into a room with strong secret-project vibes. The word “ASTRA” was emblazoned in giant letters across one wall. Xu’s dog, Charlie, the project’s de facto mascot, roamed between desks where researchers and engineers were busy constructing a product that Google is betting its future on.
