Google Revolutionizes Gemini with Video AI and Immersive Interface

⚡

Key Takeaways

1Google introduces Neural Expressive, an interactive design for Gemini, enhancing the user experience with smooth visuals and animations.

2The Gemini 3.5 Flash model promises fast and efficient execution of complex tasks, outperforming its predecessors in performance.

3Gemini Omni enables the creation of cinematic videos from various media, integrating a digital watermark for authenticity.

💡Why it matters — These innovations strengthen Google's position in AI, providing powerful tools for users and content creators.

During its annual Google I/O keynote, Google unveiled a series of impressive updates for its Gemini application, showcasing innovations in artificial intelligence and interface design. These new features promise to radically transform the user experience.

Neural Expressive: a reinvented interface

Google introduced Neural Expressive, a new visual language that redefines the look of Gemini for the AI era. This modern design incorporates smooth animations, revamped typography, and haptic feedback, allowing the assistant to provide enriched responses with visuals, interactive timelines, and narrated videos, depending on the content of the query. The Gemini Live experience is now integrated directly into the main app, facilitating a seamless transition between text and voice conversation. The microphone has been enhanced to capture hesitations and rephrasing, enabling smoother interaction. Neural Expressive is already available on the web, Android, and iOS for all users.

Gemini 3.5 Flash: a next-generation model

The Gemini 3.5 Flash model marks a significant advancement in executing complex workflows. Presented as Google's most powerful coding and action model to date, it surpasses the previous model, Gemini 3.1 Pro, with high scores on benchmarks such as Terminal-Bench 2.1 (76.2%) and MCP Atlas (83.6%). In terms of generation speed, Google claims it is four times faster than comparable models. Thanks to the Antigravity system, it can orchestrate sub-agents in parallel, allowing tasks that previously took days for a developer or weeks for an auditor to be completed in record time, often at less than half the cost of other leading models. This model is now the standard for the Gemini application and serves as the foundation for Gemini Spark. A Pro version is being finalized, with deployment expected next month.

Gemini Omni: AI video at your fingertips

With Gemini Omni, Google offers a powerful tool for video creation and editing. This multimodal model uses Gemini's reasoning to generate coherent scenes from text, images, videos, or audio. It allows users to edit an existing video through successive natural language instructions while maintaining character and scene consistency from one turn to the next. Users can transform the environment of a video, whether it's the setting, visual style, or effects, or modify the filmed action via a prompt. Gemini Omni also enables the generation of videos from mixed references, such as a character image, ambient audio, or a reference motion video, combined into a single coherent rendering. Additionally, it is possible to create explanatory videos from a short prompt, with an understanding of physical laws and cultural or scientific context. Finally, Gemini Omni offers the ability to produce a digital avatar that resembles the user's image and voice to generate personalized videos. All videos produced with Omni incorporate the digital watermark SynthID from Google DeepMind, verifiable from the Gemini app, Gemini in Chrome, and Google Search.

Gemini Spark: a personal agent 24/7

Gemini Spark is presented by Google as a personal AI agent that operates in the background, even when the screen is locked. It relies on Gemini 3.5 and integrates with Workspace tools (Gmail, Docs, Slides) as well as third-party applications via MCP connections. Among the first announced partners are Canva, OpenTable, and Instacart. Specifically, Spark can automate recurring tasks, such as analyzing bank statements or extracting deadlines from school emails, create complete workflows from meeting notes, or autonomously draft Google Docs. Google specifies that the agent requests confirmation before executing sensitive actions, such as sending an email or making a purchase.

Daily Brief: a tailored morning recap

The Daily Brief is an agent designed to generate a personalized summary every morning. It aggregates data from Gmail and Google Calendar to produce an action-oriented daily briefing, including urgent messages, upcoming events, and suggested next steps. Users can refine the agent's priorities over time through a simple feedback system, using a thumbs up or down. Daily Brief builds on the work of the CC project, an experiment launched via Google Labs, but the feature also resembles Pulse, a similar option unveiled by ChatGPT last September.

How to access the new Gemini features

The five announcements from Google I/O 2026 are not all available at the same pace or for the same profiles. Neural Expressive is available today for all users on the web, Android, and iOS. Gemini 3.5 Flash is also accessible today for all users via the Gemini app and AI Mode in Google Search. Gemini Omni Flash is available today for Google AI Plus, Pro, and Ultra subscribers, as well as for free on YouTube Shorts. Gemini Spark is being rolled out to trusted testers this week, with a beta expected for Google AI Ultra subscribers in the United States next week. Finally, Daily Brief is being rolled out for Google AI Plus, Pro, and Ultra subscribers in the United States.