The 5 biggest changes coming to Gemini

Google used its annual I/O developer conference on May 19, 2026, to outline a broad overhaul of the Gemini app. The company introduced a new flagship model, a redesigned interface, a video-focused multimodal system, an always-on AI agent, and a daily briefing tool. The changes point to a single strategic direction: Gemini is moving from a chatbot that answers prompts to a platform that can generate interactive media, act in the background, and organize personal information across connected apps. Google says the updates will reach different products and subscriber tiers on different schedules.

Key facts at a glance

• Gemini 3.5 Flash is coming to the Gemini app and AI Mode in Search. Google describes it as faster and more efficient than leading models, with richer interactive web interfaces and graphics. It also beats Gemini 3.1 Pro on coding and agentic benchmarks, according to the company.
• The Gemini app is receiving a Neural Expressive redesign with smoother animations, more vibrant colors, new typography, and haptic feedback. Responses will prioritize key information and add imagery, interactive timelines, narrated videos, and dynamic graphics.
• Gemini Omni Flash is a new multimodal model that initially generates videos from images, audio, video, and text. Google says it can create realistic scenes with more accurate physics and reason about what comes next. An Avatars feature can create videos with a digital version of a user and their own voice.
• Gemini Spark is an always-on AI agent that works in the background. It can send emails, scan credit card statements for hidden subscription fees, summarize meeting notes, connect to Workspace and third-party apps, and access local files on macOS.
• Gemini Daily Brief gathers information from connected apps to produce a daily rundown with upcoming Calendar events and Gmail updates. It prioritizes items based on user goals and accepts thumbs-up or thumbs-down feedback.

A faster and more capable Gemini 3.5 Flash heads to the Gemini app

The first major change is the arrival of Gemini 3.5 Flash in the Gemini app and AI Mode in Search. Google positions the model as the latest version of its flagship AI system, but the more important message is about efficiency. The company says Gemini 3.5 Flash is faster and more efficient than other leading AI models while offering richer, more interactive web user interfaces and graphics. That combination matters because consumer AI products increasingly compete on responsiveness and presentation, not just raw text generation. A model that can return answers quickly and render useful visual elements can make the app feel less like a search box and more like a workspace.

Google also says the model represents a major leap forward for coding useful AI agents, beating Gemini 3.1 Pro on coding and agentic benchmarks. That claim is significant because agentic benchmarks test whether a model can plan, use tools, and carry out multi-step tasks rather than simply produce plausible prose. If Gemini 3.5 Flash performs well in those areas, it could support features that require the model to act on behalf of a user across apps and services. The model is only the first in the Gemini 3.5 family. Google plans to launch Gemini 3.5 Pro next month, which suggests the company is separating speed and cost from higher-end capability rather than relying on a single model for every task.

The rollout also shows how Google is distributing AI across its products. Gemini 3.5 Flash is not limited to a standalone app; it is also coming to AI Mode in Search. That placement could expose the model to a much larger audience and make Gemini-style interactions part of the core search experience. For users, the practical difference may be faster answers, better formatting, and more capable assistance with tasks that involve code or multi-step planning. For Google, the goal is to make the model useful in enough contexts that people encounter it without deliberately opening a separate app.

Gemini gets a Neural Expressive redesign

Alongside the new model, Google is giving the Gemini app a makeover through a design language called Neural Expressive. The redesign includes smoother animations, more vibrant colors, new typography, and haptic feedback when users tap buttons. Those details may sound cosmetic, but they matter for a product that is trying to establish a distinct identity in a crowded AI market. A more responsive interface can make complex model outputs feel more approachable, while haptic feedback and animation can signal that the app is working even when a request takes time.

Google is also reformatting Gemini's responses. Instead of returning a wall of text, the app will put the most important information at the top and include imagery, interactive timelines, narrated videos, and dynamic graphics. That shift reflects a broader change in how AI answers are presented. Text remains the foundation, but the output can become a visual briefing, a timeline, or an interactive explainer. For tasks such as research, planning, and learning, the format of an answer can be as important as the answer itself. A clear summary followed by visual context may be easier to scan than a long paragraph, especially on mobile devices.

The redesign also makes it easier to switch from typing text in the Gemini app to the voice-enabled Gemini Live mode. Voice interaction has become a central part of AI assistants because it supports hands-free use and more natural back-and-forth conversation. Google says Gemini Live will soon have new regional dialects, which could improve accessibility for users who prefer to speak in local accents or language variants. The Neural Expressive redesign is rolling out across the Gemini app on the web, Android, and iOS, so the update is intended to be consistent across platforms rather than limited to a single device.

Gemini Omni Flash can create videos using any input

Google is also updating the Gemini app with a new kind of model that it says will eventually create anything from any input. The first model in the family, Gemini Omni Flash, will initially allow users to generate videos using a combination of images, audio, video, and text. That makes it a multimodal creation tool rather than a text-to-video system alone. Users could supply reference images, audio cues, existing video, and written instructions, then ask the model to produce a new clip. Google says Gemini Omni can create realistic-looking scenes with more accurate physics and can reason about what comes next, which is important for generating video that holds together over time.

The company is also adding an Avatars feature that can create videos with a digital version of the user, using the user's own voice. That feature raises obvious questions about consent, identity, and misuse, but it also points to a practical use case: creating personalized video messages or presentations without recording a new take each time. Google says users will be able to edit Gemini Omni's output through conversations in the Gemini app. That conversational editing model is becoming common in creative tools because it lets users make changes by describing them rather than navigating a complex timeline editor.

Gemini Omni Flash is rolling out to Google AI Plus, Pro, and Ultra subscribers in the Gemini app and Google Flow. It will also be available for free on YouTube Shorts and the YouTube Create App. The distribution plan is notable because it puts advanced video generation in front of both paying subscribers and casual creators. YouTube Shorts is a mass-market surface with a constant demand for new clips, while Google Flow is aimed at creative workflows. By placing the same model family across those products, Google can test how people use generated video at different levels of sophistication and collect feedback from a much broader audience.

Gemini Spark is Google's take on OpenClaw

Google is launching Gemini Spark, an always-on AI agent that performs work in the background while the user completes other tasks. The company compares it to OpenClaw, an open-source AI agent platform that attracted attention across the tech industry earlier in the year. The comparison is useful because it highlights the shift from conversational AI to agentic AI. A chatbot waits for a prompt and returns an answer. An agent can be given a goal, break it into steps, use connected tools, and continue working without constant supervision.

Google says Gemini Spark can send emails, scan monthly credit card statements for hidden subscription fees, and create a summary of meeting notes. Those examples are deliberately mundane. They are the kind of small administrative tasks that consume time but do not require deep expertise. That is often where agents can deliver the most immediate value: not by replacing complex professional judgment, but by handling repetitive work that people would rather not do themselves. The credit card example is especially telling because it requires the agent to read documents, identify patterns, and surface potential savings. It is a task that combines document understanding with practical reasoning.

Gemini Spark runs on Gemini 3.5 Flash and can connect to Workspace apps such as Docs, Sheets, and Slides, along with third-party services such as Canva, Instacart, and OpenTable. The AI agent can also tap into local files on macOS using the Gemini app. That range of connections is central to the product. An agent is only as useful as the tools it can reach, and Google is trying to make Gemini Spark a hub that can operate across productivity software, design tools, shopping services, and restaurant reservations. The macOS file access is particularly important because it lets the agent work with documents and data that may never have been uploaded to a cloud service.

Gemini Spark is rolling out to trusted testers this week, followed by a launch in beta for US-based Google AI Ultra subscribers next week. The staged rollout reflects the risks involved in always-on agents. An agent that can send emails, read statements, and access local files needs strong permission controls, clear audit trails, and reliable ways to stop or correct mistakes. Trusted testers can surface problems before a wider release, and the Ultra subscriber tier gives Google a smaller group of highly engaged users for the beta. The success of Gemini Spark will depend less on demo performance than on whether users feel comfortable letting it act on their behalf.

Gemini Daily Brief offers a rundown of your day

Google is bringing another AI-powered agent to the Gemini app, called Daily Brief. It gathers information from the apps a user has connected to Gemini, allowing it to put together a summary of the day with upcoming events from Calendar and updates received in Gmail. The feature is designed to reduce the effort of checking multiple apps first thing in the morning. Instead of opening a calendar and an inbox separately, a user can get a single briefing that combines scheduling information with important messages.

Google says Daily Brief will organize and prioritize its rundowns based on user goals, and it will accept feedback through thumbs-up or thumbs-down ratings. That feedback loop is important because prioritization is highly personal. One user may care most about meetings and deadlines, while another may want urgent emails or family reminders at the top. Over time, a daily briefing tool that learns from explicit ratings can become more useful than a static digest. The challenge is to make the prioritization transparent enough that users understand why an item appears and can correct the system when it gets something wrong.

Daily Brief is coming to Google AI Plus, Pro, and Ultra subscribers in the US starting Tuesday. The launch is limited by both geography and subscription tier, which is consistent with how Google is rolling out many of its advanced Gemini features. The company is using paid tiers to manage demand and test features with users who have already opted into its AI subscription products. Taken together with the other I/O announcements, Daily Brief shows how Google wants Gemini to become a personal assistant that not only answers questions but also organizes information, creates media, and completes tasks across the apps people already use.

Source:The Verge News