Google I/O 2026: All about Gemini 3.5, Spark, Omni models, and revamped app

At Google I/O 2026, Google introduced Gemini 3.5 AI models, Gemini Omni for video generation, new agentic experiences with Gemini Spark, and updates across the Gemini app

Google kicked off its annual developer conference, Google I/O 2026, on May 19 with a keynote address focusing on Gemini and its expanding AI ecosystem. The company announced a range of updates spanning new Gemini AI models, agentic experiences, and several changes to the Gemini app.

 

Google introduced Gemini 3.5, its latest family of AI models designed to improve speed, coding, automation, and long-task handling. Google also unveiled Gemini Omni, a new AI model built for cinematic video generation and editing using text prompts, images, and video clips.

 

Alongside the new models, Google introduced Gemini Spark, a proactive AI agent that can automate workflows and continue running tasks in the background. This is in addition to several new features coming to the Gemini app, such as Daily Brief for personalised overviews, a new UI design, updated voice features, and more.

 

 

Gemini’s new models

 

Google has announced Gemini 3.5, its newest family of AI models focused on improving both intelligence and real-world task handling. The company said the new models are designed to power more capable AI agents that can manage complex and longer tasks more effectively.

 

Google is starting the rollout with Gemini 3.5 Flash, which it says delivers stronger performance for coding, automation, and agent-based tasks while also being fast and responsive. According to Google, Gemini 3.5 Flash is now available globally through the Gemini app and AI Mode in Google Search. Google also confirmed that Gemini 3.5 Flash powers several of the new AI features announced at I/O, including the Gemini Spark agent.

 

The company said it is also working on Gemini 3.5 Pro, a more advanced version of the AI model. Gemini 3.5 Pro is already being tested internally and is expected to roll out publicly next month.

 

Another major announcement at the event was Gemini Omni, a new AI model focused on video creation and editing. Google said the tool can combine text prompts, images, and video clips to generate cinematic-quality video content directly inside Gemini. Google said the model combines an understanding of physics with Gemini’s broader knowledge of history, science, and cultural context to create videos that feel more logical and story-driven rather than just visually realistic.

 

Google announced that the first model in the Omni lineup, called Gemini Omni Flash, is rolling out globally starting today for Google AI Plus, Pro, and Ultra subscribers through the Gemini app and Google Flow. Google also confirmed that Gemini Omni Flash will become available for users on YouTube Shorts and the YouTube Create app at no additional cost starting this week.

 

Agentic Gemini experience

 

Google has also introduced Gemini Spark, a cloud-based AI agent designed to manage tasks on behalf of users. Google describes Spark as a major evolution for Gemini, shifting it from a tool that simply answers questions into a system that can perform ongoing work independently under user supervision.

 

According to Google, Spark runs continuously in the background, meaning it can continue completing tasks even after a user locks their phone or closes their laptop. The feature is deeply integrated with Google Workspace services such as Gmail, Calendar, Docs, and Slides. Google noted that Spark is powered by Gemini 3.5 and uses its “Antigravity harness” system. Essentially, Antigravity is Google’s agent-first platform that allows developers to build, manage, and deploy AI agents that work across different Google tools and services.

 

The company shared several examples of how Spark can be used. Users can set recurring automation tasks, such as monitoring monthly credit card statements to identify hidden subscription charges. Spark can also monitor school-related emails, extract important deadlines, and send summarised updates to family members.

 

In work scenarios, Google said the agent can analyse meeting notes spread across emails and chats, create Google Docs summarising key points, and even draft project kickoff emails automatically.

 

Google also announced new MCP integrations with companies including Canva, OpenTable, and Instacart. According to Google, Spark will soon be able to use these integrations to complete tasks directly for users. The company added that more capabilities are planned in the coming months, including emailing Spark, texting Spark, creating custom sub-agents, and allowing it to control browsers.

 

Google has also previewed Android Halo, a new feature that lets users track what their AI agent is doing in real time directly from the top of the phone screen. The company said Halo will show live updates when an agent is working on a task, entering live mode, or sending messages, without interrupting what the user is currently doing.

 

The announcement builds on updates revealed earlier at The Android Show 2026, where Google introduced Gemini Intelligence as part of the upcoming Android 17 experience. Android Halo will launch later this year and will work with Gemini Spark and other supported AI agents. 

 

Gemini app updates

 

From now on, Gemini 3.5 Flash will power the Gemini app by default. These are the changes coming to the Gemini app:

 

Daily Brief: Google has introduced a new feature called Daily Brief, which works as a proactive morning summary inside the Gemini app. The feature collects updates from connected Google services like Gmail and Calendar and presents them in a single organised briefing.

 

According to Google, Daily Brief can identify urgent emails, upcoming meetings, reminders, and follow-up details while also suggesting next actions based on a user’s priorities. The feature is similar to Samsung’s “Now Brief” experience on Galaxy phones, which also delivers personalised summaries throughout the day.

 

‘Neural Expressive’ interface: Google has redesigned the Gemini app with a new visual language called “Neural Expressive.” The updated interface introduces fluid animations, brighter colours, new typography, and haptic feedback to make interactions feel more dynamic and less text-heavy. Google said the redesign is intended to create a more natural AI experience across Android, iOS, and the web.

 

Gemini Live: Google has integrated Gemini Live directly into the main Gemini experience, allowing users to switch between typing and speaking without changing modes. The company also redesigned the microphone system so users can pause, rethink ideas, or continue speaking naturally without Gemini interrupting them midway. Google also plans to add regional dialect support so users can choose voices that better match their accent and speaking style.

 

Gemini responses become more visual and interactive: Instead of only showing long blocks of text, Gemini can now generate richer responses using interactive timelines, visuals, narrated clips, and dynamic graphics. According to Google, this is designed to make information easier to understand and more engaging for users.

 

Gemini app for macOS: Google also announced major updates for the Gemini app on macOS. Later this summer, Gemini Spark will arrive on the desktop app, allowing Gemini to work with local files and automate workflows directly on a user’s computer.

 

The macOS app is also getting new voice features that can convert natural speech into properly formatted text while understanding context from the screen. Google says the system is designed to ignore filler words and pauses so users can quickly create drafts and notes more naturally.