Google I/O 2026: May 19-20 — Gemini 4, Android 17, What to Expect

Google I/O 2026 is confirmed for May 19-20 at Shoreline Amphitheatre, Mountain View, with the developer keynote at 10am PT on May 19. This is the most anticipated Google developer event in years — the session list already published reveals Gemini 4, Android 17, and a major shift in how Google thinks about agentic AI tooling.

Here is what the confirmed sessions tell us before the keynote.

Gemini 4: What the Session List Reveals

Google has not officially confirmed "Gemini 4" as a name, but the I/O session list contains sessions titled around next-generation Gemini capabilities that go significantly beyond Gemini 3.1 Ultra.

The confirmed session themes point to three capability areas:

Context window expansion: Gemini 3.1 Ultra already has 2 million tokens. The I/O sessions around "Gemini long context" suggest the next model pushes further — potentially toward 10 million tokens. At that scale, a developer could feed an entire large codebase (1M+ lines) as context in a single API call.

Native multimodal: Unlike GPT-4o and Claude Opus 4.7 which process modalities sequentially, the Gemini 4 session descriptions emphasise native text/image/audio/video processing without transcription preprocessing. The practical difference: audio and video analysis without converting to text first, preserving tonal and temporal information that gets lost in transcription.

Agentic coding focus: Three separate I/O sessions cover agentic coding workflows — long-running code agents, multi-step software engineering tasks, and integration with Google's development tools. This is a direct response to Claude Code and OpenAI's Codex, positioning Gemini as the AI layer underneath Google's entire developer toolchain.

Android 17: The Platform Merge

Android 17 is confirmed at I/O under the codename "Adaptive Everywhere." The significance: it is not just an Android OS update. It is the operational merger of Android, Chrome OS, and Google's XR (extended reality) platform into a single unified system.

For developers, this is the biggest Android platform change since Android 5.0 Lollipop introduced Material Design. A unified Android/ChromeOS/XR codebase means:

Single app target: Apps built for Android 17 will run natively on phones, tablets, Chromebooks, and XR headsets without separate builds. The responsive layout system is being redesigned from the ground up to accommodate form factors from a 4-inch phone screen to a full XR immersive environment.

Chrome OS app ecosystem: Chromebooks currently run Android apps in a compatibility layer. With Android 17, the compatibility layer disappears — Chromebooks run Android 17 natively. This expands the addressable market for Android apps to every Chromebook sold.

XR developer surface: Google has not shipped a successful XR headset, but Android 17 is the infrastructure play for when they do — or when third-party OEMs do. Developers building for Android 17 are also building for Google's XR future without additional work.

The migration path from Android 16 to Android 17 will involve the most significant API changes since Android 6.0 runtime permissions. Start reading the developer preview documentation the week it drops at I/O.

Flutter, Firebase, and Google Play

The session list confirms major updates to all three:

Flutter: New rendering engine updates and performance improvements for the Android 17 unified platform. Flutter's cross-platform model becomes more valuable when a single Flutter app targets phone, tablet, Chromebook, and XR. Expect Dart language updates and new widget libraries for the expanded form factor range.

Firebase: AI features integrated directly into Firebase — Gemini-powered functions, AI-native authentication patterns, and real-time model inference from Firebase functions. The direction is clear: Firebase becomes the backend for AI-powered mobile apps, not just traditional database/auth/storage apps.

Google Play: New AI model distribution capabilities — developers will be able to distribute on-device AI models through Google Play with the same infrastructure as app updates. This is the infrastructure for running smaller Gemini models on-device in Android apps without API calls.

The Veo 3 Question

Veo, Google's text-to-video model, is expected to have a significant I/O update. The current Veo 2 generates high-quality video from text prompts. I/O session descriptions hint at Veo 3 capabilities including audio generation alongside video and longer video sequences.

For developers: Veo API access has been limited since launch. If I/O opens Veo 3 to general developer API access, it is the most significant new capability for content and media applications since the image generation API wave of 2023.

What Developers Should Prepare For

Before I/O (now through May 19):

• Review Android 17 developer preview when it drops (likely released at I/O or shortly before)
• Audit your app's layout code for Adaptive Everywhere compatibility — the form factor changes will break fixed-size layouts
• Check Flutter and Firebase changelog for breaking API changes

At I/O (May 19-20):

• Watch the developer keynote live — Gemini 4 API access announcements typically include immediate developer access or a waitlist that fills fast
• Register for extended sessions on agentic coding if you are building AI-powered developer tools

Post I/O:

• Gemini 4 API migration from 3.1 will likely require prompt engineering adjustments for the new context window behaviour
• Android 17 target API level update will be required for Play Store by a deadline announced at I/O (historically 12-18 months after release)

Key Takeaways

• Google I/O 2026 is May 19-20, keynote 10am PT May 19 at Shoreline Amphitheatre — register now, developer sessions fill immediately after keynote
• Gemini 4 expected: 10M+ token context, native multimodal (no transcription preprocessing), three dedicated agentic coding sessions — direct response to Claude Code and OpenAI Codex
• Android 17 "Adaptive Everywhere": merges Android, Chrome OS, and XR into single platform — biggest Android developer change since Material Design; single app target for phone/tablet/Chromebook/XR
• Flutter, Firebase, Google Play all major updates: Flutter for unified platform, Firebase AI functions, on-device model distribution through Play Store
• Veo 3 likely: general developer API access to text-to-video + audio generation is the wildcard announcement to watch

For current AI model comparison including Gemini 3.1, read Claude vs ChatGPT, Gemini and Grok 2026. For live API pricing across providers, see LLM API Pricing. For the Claude Opus 4.7 release that shipped today, read Anthropic Launches Claude Opus 4.7 and Claude Design.