Google I/O 2026: Gemini 3.5 and Gemini Omni Are the Main Story

Google I/O 2026 made one thing clear: Google wants Gemini to become more than a chatbot, more than a search feature, and more than a model family. The biggest theme of the event was the move toward an agentic Gemini ecosystem, with Gemini 3.5 powering action and Gemini Omni expanding what users can create from multimodal input.

The announcements covered Search, Android, developer tools, Google Flow, YouTube, subscriptions, Workspace, shopping, and personal agents. But the central narrative was easy to identify. Gemini 3.5 Flash is Google's new engine for fast, capable agents and coding workflows. Gemini Omni is Google's new creative model family for generating and editing media, starting with video.

Together, they show where Google is taking AI in 2026: from answering prompts to completing tasks, from static responses to interactive interfaces, and from isolated AI tools to models embedded across Google's major products.

The Big Picture From Google I/O 2026

Google framed I/O 2026 as the arrival of a more action-oriented AI era. Instead of presenting Gemini only as a conversational assistant, the company emphasized agents, tools, multimodal creation, persistent workflows, and deeper product integration.

The official I/O roundup highlighted two new model directions:

• Gemini 3.5, beginning with Gemini 3.5 Flash, a model designed for frontier intelligence, speed, coding, and long-horizon agentic workflows.
• Gemini Omni, a new model family that can create from multiple kinds of input, beginning with video generation and video editing through Gemini Omni Flash.

This matters because these two model families serve different but connected purposes. Gemini 3.5 is the action layer. It is built to reason, plan, use tools, work with agents, and support products like AI Mode in Search, Gemini Spark, Google Antigravity, and Managed Agents in the Gemini API. Gemini Omni is the creation layer. It is built to turn text, images, video, and audio references into coherent media outputs.

In other words, Google is not only trying to win model benchmarks. It is trying to build a complete AI operating layer across productivity, software development, search, shopping, and creative work.

Gemini 3.5 Flash: The First Model in Google's New 3.5 Family

Gemini 3.5 Flash is the first model in the Gemini 3.5 series. Google introduced it on May 19, 2026, and positioned it as a model that combines frontier intelligence with action.

That phrase is important. Google is no longer describing Flash only as the fast, lightweight option. Gemini 3.5 Flash is being presented as a serious agentic and coding model that can handle complex, multi-step work while still keeping the speed expectations of the Flash line.

According to Google, Gemini 3.5 Flash is available through:

• The Gemini app
• AI Mode in Google Search
• Google Antigravity
• The Gemini API in Google AI Studio
• Android Studio
• Gemini Enterprise Agent Platform
• Gemini Enterprise

Google also said Gemini 3.5 Pro is already being used internally and is expected to roll out after Flash. That makes Gemini 3.5 Flash both a product launch and a preview of the broader Gemini 3.5 strategy.

Why Gemini 3.5 Flash Matters

The most important claim around Gemini 3.5 Flash is not simply that it is faster. The real claim is that it can support long-horizon agentic work.

Traditional AI assistants often work best on short tasks: answer a question, summarize a document, write a function, or generate a block of text. Agentic systems are different. They need to inspect context, make plans, call tools, write files, review results, and continue across multiple steps. That is the kind of work Google is targeting with Gemini 3.5 Flash.

Google says Gemini 3.5 Flash outperforms Gemini 3.1 Pro on several coding and agentic benchmarks, including Terminal-Bench 2.1, GDPval-AA, and MCP Atlas. It also says the model is significantly faster than other frontier models in output speed. Benchmarks should not be treated as a complete picture of real-world performance, but they do show the direction of the release: Google wants Flash to become a default model for practical AI agents, not just quick chat responses.

For developers, this is one of the biggest takeaways from I/O 2026. Gemini 3.5 Flash is designed for tasks such as maintaining codebases, building applications, automating repeated workflows, handling tool calls, and coordinating subagents. That makes it especially relevant for software teams, AI coding products, enterprise automation, and internal productivity tools.

Gemini 3.5 and the Agentic Search Strategy

Gemini 3.5 Flash is also now the default model for AI Mode in Google Search globally. That is a major distribution move. Search is one of Google's most important products, and making Gemini 3.5 Flash the default model inside AI Mode means Google is pushing the new model into a high-usage consumer surface immediately.

At I/O 2026, Google described a broader AI Search shift:

• AI Mode has surpassed 1 billion monthly users.
• Google is bringing AI Overviews and AI Mode closer together.
• Search can use text, images, files, videos, and Chrome tabs as input.
• Google is introducing information agents that can monitor topics and projects in the background.
• Search will eventually help build custom mini-app experiences for ongoing tasks.

This is where Gemini 3.5 Flash becomes strategically important. Search is no longer just a place to retrieve links. Google wants it to become a place where users can ask, continue, compare, monitor, and act. A model built for fast agentic workflows is central to that plan.

Gemini Spark: A Personal Agent Built on Gemini 3.5

Another major I/O 2026 announcement was Gemini Spark, Google's personal AI agent. Spark is designed to operate in the background, help users navigate digital tasks, and take action under user direction.

Google said Spark runs on Gemini 3.5 and is built on the Google Antigravity platform. The first rollout is cautious: Google described Spark as early in its product journey, starting with trusted testers and then a beta for Google AI Ultra subscribers in the United States.

Spark is important because it shows how Google thinks Gemini 3.5 will reach everyday users. The model is not only for developers writing code. It is also intended to power agents that can summarize daily information, manage tasks, work across apps, and eventually handle more complex user-authorized actions.

The key point is supervision. Google repeatedly emphasizes that these agents operate under user direction and should check with users before major actions. That framing will matter as personal AI agents become more capable and more deeply connected to user accounts, files, calendars, messages, and payment flows.

Google Antigravity and Developer Tools

For developers, one of the clearest I/O 2026 messages was that Google is putting Gemini 3.5 Flash at the center of its agentic development stack.

Google announced updates around:

• Google Antigravity 2.0, a standalone desktop application for orchestrating agents.
• Antigravity CLI, a terminal-first interface for creating and running agents.
• Antigravity SDK, which gives developers programmatic access to the agent harness.
• Managed Agents in the Gemini API, where a single API call can start an agent in an isolated Linux environment.
• Google AI Studio upgrades, including Android app generation and better handoff into Antigravity.

Managed Agents may be especially important. Google says they are powered by the Antigravity harness and Gemini 3.5 Flash. The idea is to let developers create agents that can reason, use tools, execute code, and preserve state across interactions without building all the infrastructure themselves.

This makes Gemini 3.5 Flash more than a model endpoint. It becomes part of a developer platform for building agentic applications.

Gemini Omni: Google's New Creative Model Family

If Gemini 3.5 Flash is about action, Gemini Omni is about creation.

Google introduced Gemini Omni as a new model family that can create from any input, starting with video. The first release is Gemini Omni Flash, which is rolling out to the Gemini app, Google Flow, and YouTube Shorts-related creation tools.

Gemini Omni is built around a simple but ambitious promise: users should be able to combine text, images, video, and audio references, then generate or edit high-quality video through natural language. Over time, Google says Omni will support more output modalities, including image and audio.

That makes Omni different from a normal text-to-video model. Google is presenting it as a broader multimodal creation system. It can use references, preserve characters, understand scenes, edit across multiple turns, and generate content grounded in Gemini's world knowledge.

What Gemini Omni Can Do

Gemini Omni is designed for conversational video creation and editing. Instead of using a traditional editing timeline, users can describe what they want changed.

The main capabilities include:

• Generating video from text prompts.
• Editing existing videos with natural language.
• Using image, text, video, and audio references.
• Preserving character consistency across scenes.
• Applying visual styles, camera changes, and motion references.
• Creating videos with more realistic physics and scene continuity.
• Supporting personal digital avatars through Google's avatar tools.

Google also emphasized that Omni has a stronger intuitive understanding of forces such as gravity, kinetic energy, and fluid dynamics. The goal is not only to make clips look realistic, but to make the action inside them feel more physically coherent.

This is one of the reasons Gemini Omni was one of the biggest announcements at I/O 2026. Video generation is becoming a major competitive area in AI, but practical video editing is still difficult for most users. If Omni can make high-quality video editing conversational and reference-driven, it could become a major creative tool for everyday users, marketers, filmmakers, educators, and social media creators.

Gemini Omni in the Gemini App, Google Flow, and YouTube

Google is launching Gemini Omni Flash across several surfaces rather than keeping it in a single lab product.

In the Gemini app, Omni gives users a conversational way to create and edit videos. A user can upload a photo or video, apply templates, change backgrounds, add cinematic movement, or create a personalized avatar.

In Google Flow, Omni becomes part of Google's creative studio. Flow users can use real-world references, iterate through prompts, preserve characters across scenes, and combine generated assets into broader creative projects. Google also announced Flow Agent, which can help with multi-step creative work such as brainstorming, generating variations, batch editing, and organizing assets.

In YouTube, Gemini Omni appears in Shorts Remix and the YouTube Create app for eligible users. This is a particularly important distribution point because Shorts creators need fast, lightweight tools that can remix existing content, add references, and produce shareable results quickly.

The pattern is familiar: Google is not treating Omni as a standalone demo. It is embedding the model into products where creation already happens.

Safety, Watermarking, and Content Verification

Generative video raises obvious questions about authenticity, misuse, and trust. Google addressed this by tying Gemini Omni to SynthID watermarking and content verification tools.

Google says videos created with Omni include an imperceptible SynthID digital watermark. It also said users can verify content through the Gemini app, Gemini in Chrome, and Search. At I/O 2026, Google also discussed broader support for C2PA Content Credentials, which are intended to help people understand whether content is original, modified, or AI-generated.

These systems will not solve every trust problem around AI media, but they are important because video generation is becoming easier and more realistic. As tools like Gemini Omni reach consumer apps and YouTube workflows, verification will become a necessary part of the product experience, not a separate policy footnote.

Why Gemini 3.5 and Gemini Omni Are the Main Story

Google I/O 2026 had many announcements, but Gemini 3.5 and Gemini Omni stand out because they define the two biggest directions for Google's AI strategy.

Gemini 3.5 is about doing. It powers agents, coding tools, Search experiences, app building, enterprise workflows, and personal AI assistants. It is Google's attempt to make AI more useful by giving it the speed, planning ability, and tool support needed to complete real tasks.

Gemini Omni is about creating. It brings Gemini's multimodal reasoning into video generation and editing, with a roadmap toward broader input and output combinations. It is Google's attempt to make creative work more conversational, more reference-aware, and more accessible.

Put together, these models show that Google is no longer positioning Gemini as one product. Gemini is becoming a platform layer:

• In Search, it helps users ask deeper questions and build custom experiences.
• In the Gemini app, it becomes more proactive and agentic.
• In Antigravity, it powers coding agents and developer workflows.
• In the Gemini API, it supports managed agent infrastructure.
• In Flow and YouTube, it becomes a creative engine.
• In enterprise products, it automates complex operational work.

That is why Gemini 3.5 and Gemini Omni were the biggest storyline at I/O 2026. They are not isolated releases. They are the foundation for Google's next phase of AI products.

What This Means for Users and Developers

For everyday users, the practical impact is that Gemini will become more present across Google products. Search will feel more conversational and agentic. The Gemini app will become more proactive. Video creation tools will become more accessible through Gemini Omni. Personal agents like Spark will start to test how much work users are willing to delegate to AI.

For developers, Gemini 3.5 Flash is the more immediate story. The model is generally available through developer surfaces, and Google's Antigravity and Managed Agents announcements suggest a serious push into agent-building infrastructure. Developers should watch how pricing, rate limits, tool reliability, context handling, and stateful agent environments mature after launch.

For creators, Gemini Omni is the announcement to watch. If its editing workflow works well in real production settings, it could reduce the gap between idea, footage, and finished video. The most important test will be whether it can preserve character identity, motion, style, timing, and scene continuity across repeated edits.

For businesses, the story is ecosystem integration. Google is connecting Gemini to Search, Workspace, Commerce, YouTube, Android, enterprise tools, and developer platforms. The value will come less from one model feature and more from how reliably these tools work together.

Final Thoughts

Google I/O 2026 was an AI-heavy event, but it was not only about bigger models. It was about making AI more operational.

Gemini 3.5 Flash gives Google a fast agentic model for coding, Search, personal agents, and enterprise automation. Gemini Omni gives Google a creative model family for multimodal video generation and editing. One is built to act. The other is built to create. Together, they form the clearest signal from I/O 2026: Google wants Gemini to become the intelligence layer across everyday digital work.

The next question is execution. If Gemini 3.5 agents are reliable enough and Gemini Omni's video tools are consistent enough, Google could turn I/O 2026 into a major turning point for its AI ecosystem. If not, the announcements will still be remembered as ambitious, but the market will wait for proof in daily use.

Sources: Google I/O 2026 official roundup, Gemini 3.5 announcement, Gemini Omni announcement, Gemini app updates, and I/O 2026 developer highlights.

FAQ

What was the biggest AI announcement at Google I/O 2026?

The biggest AI storyline was the combination of Gemini 3.5 and Gemini Omni. Gemini 3.5 Flash powers agentic workflows, coding, Search, and developer tools, while Gemini Omni introduces multimodal creation and video editing.

What is Gemini 3.5 Flash?

Gemini 3.5 Flash is the first model in Google's Gemini 3.5 family. It is designed for fast, capable agentic work, including coding, tool use, long-horizon tasks, and AI-powered product experiences.

Is Gemini 3.5 Flash available now?

Yes. Google announced Gemini 3.5 Flash on May 19, 2026, and said it is available through the Gemini app, AI Mode in Search, Google Antigravity, the Gemini API in Google AI Studio, Android Studio, and enterprise Gemini products.

What is Gemini Omni?

Gemini Omni is Google's new multimodal creative model family. It can create from different kinds of input, starting with video generation and video editing through Gemini Omni Flash.

What can Gemini Omni Flash create?

Gemini Omni Flash starts with video. It can generate and edit videos using text prompts, images, video references, and supported audio references. Google says future Omni models will support more output types, including image and audio.

Where is Gemini Omni Flash available?

Google says Gemini Omni Flash is rolling out through the Gemini app and Google Flow for Google AI Plus, Pro, and Ultra subscribers globally. It is also available in YouTube Shorts Remix and the YouTube Create app for eligible users.

How is Gemini 3.5 connected to Gemini Spark?

Gemini Spark is Google's personal AI agent, and Google says it runs on Gemini 3.5. Spark is designed to operate in the background under user direction and help manage digital tasks.

Why does Gemini 3.5 matter for developers?

Gemini 3.5 Flash is central to Google's developer strategy because it powers Antigravity, Managed Agents in the Gemini API, and agentic workflows that can reason, use tools, execute code, and maintain state.

Does Gemini Omni include watermarking?

Yes. Google says videos created with Omni include SynthID digital watermarking, and the company is also expanding verification tools through Gemini, Chrome, Search, and C2PA Content Credentials.

Is Gemini 3.5 Pro available?

Not yet for general public use at the time of the I/O 2026 announcement. Google said Gemini 3.5 Pro is being used internally and is expected to roll out after Gemini 3.5 Flash.