Skip to main content
Agents Take Charge

Gemini Omni and 3.5: Google unveils multimodal AI, new hardware

At its Google I/O 2026 developer conference in May, Google introduced Gemini 3.5 and Gemini Omni — a family of agentic models and a multimodal model accepting text, image, audio and video inputs — and new hardware, aiming to make AI more proactive and integrated for generative video, background agents and richer cross-device experiences.
Gemini Omni and 3.5: Google unveils multimodal AI, new hardware
Gemini Omni and 3.5: Google unveils multimodal AI, new hardware

For more than two decades, Google has invested in machine learning, research and infrastructure to integrate AI into products across healthcare, crisis response and education. The company published a roundup of its May 2026 announcements that emphasize agentic AI, new multimodal creation tools, hardware built for these experiences and initiatives applying advanced computation to science and life sciences.

Overview
May’s announcements centered on making AI more proactive and integrated. At Google I/O 2026, Google introduced Gemini 3.5 and Gemini Omni, alongside a range of product updates across Search, Android and new hardware. The month’s news also included the launch of the Google Health app, a new Fitbit device and a $10 million research effort linking quantum computing with life sciences.

Gemini Omni and creative AI
Gemini Omni was unveiled as a multimodal model that accepts images, audio, video and text as inputs to generate high-quality video grounded in real-world knowledge. The company highlighted how Omni enables creation from “any input,” marking a push into generative video and richer multimodal outputs.

Gemini Omni and 3.5: Google unveils multimodal AI, new hardware

Fonte: Gerado por AI Studio (Nano Banana)

Agentic Gemini and Search
Gemini 3.5 was presented as a family of models with enhanced action-taking abilities designed to execute complex, multi-step workflows across apps. Search is receiving agentic features that run in the background as information agents—monitoring topics, sending updates and building generative UI and interactive visuals. The Gemini app now emphasizes proactive assistance with daily briefs and background task management.

Simulations, music and coding
Project Genie combined with Street View to demonstrate browser-based, interactive 3D simulations of real places. Google Flow Music partnered with Believe to provide artists and producers an AI collaborator for songwriting and production. Agentic coding features allow Search to generate custom tools and mini apps, such as a fitness tracker that leverages live data like maps and weather.

Hardware and Android experiences
New hardware announcements include Googlebook laptops built for Gemini Intelligence and Android Halo, a phone space to manage agents and receive contextual help. Android enhancements span in-car conversational features, next-generation intelligent eyewear and deeper cross-device integrations.

Health, wearables and science initiatives
The Google Health app consolidates health and wellness features, while Fitbit Air debuts as a compact tracker offering continuous heart monitoring and advanced metrics. Research initiatives include Gemini for Science and REPLIQA, a $10 million program funding research at five universities to apply quantum computing and AI to life sciences.

Transparency and verification
Google expanded content transparency and verification tools across Search, Gemini, Chrome, Pixel and Cloud to help users understand whether online content was AI-generated.

Suggested Links
Gemini Omni: https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-omni/

Friend Caio

Life is a cycle of observing, learning, and taking action.