Today we're releasing two generative media models for developers and enterprises with strong cost-performance. 🍌 Nano Banana 2 Lite: our fastest, most cost-efficient image model in the Nano Banana family. 🌐 Gemini Omni Flash: our natively multimodal high quality, cost efficient model for video generation and conversational editing. Both models are available today in @GoogleAIStudio, the Gemini API, and the Gemini Enterprise Agent Platform.

Gemini Spark is available in Beta for Google AI Ultra subscribers aged 18 in the U.S. and this suite of new features is starting to roll out today. Learn more ↓ goo.gle/4eHGHby

Gemini Spark is coming to the @GeminiApp for macOS. Plus, we’re rolling out more updates to your 24/7 personal AI agent in the Gemini app: 🔌 Support for custom Model Context Protocol (MCP) to build a more tailored assistant 📎 Integrations with @Canva, @Dropbox, @Instacart, @OpenTable and @ZillowRentals ✅ Connections with Google Tasks and Google Keep to turn brain-dumps into action items 🗞️ Track real-time topics and events like sports, stocks and news so you don't have to hit refresh

New ways you can use Gemini Spark to make your life easier. From integration in the Gemini app for macOS to smart triggers, here are five ways we’re making Gemini Spark an even more powerful personal AI agent to help you get things done around the clock.

Catch up on more of our biggest AI updates from June 👇 goo.gle/4vQHahx

ICYMI: our June AI recap ⬇️ 🍌 Unveiled Nano Banana 2 Lite, our fastest and most cost-efficient Gemini Image model yet 💻 Introduced Gemma 4 12B, our latest open model that brings smart AI agents directly to your laptop in a single streamlined system to give you privacy without sacrificing speed ⚡ Brought Gemini Omni Flash to APIs in public previews so developers and enterprises can build custom, dynamic video workflows 🔊 Released the new Google Home Speaker, which we built with Gemini to have more natural and fluid conversations and to handle multiple requests at once 📓 Upgraded @NotebookLM with advanced reasoning, a secure cloud computer for code execution and the ability to generate charts, spreadsheets and slide decks for Google AI Ultra subscribers and specific Workspace accounts 📱Rolled out Android 17 with new features designed for multitasking, enhanced security, including lost phone biometric locking, and new personalization options 🗣️ Launched Gemini 3.5 Live Translate, which enables near-real-time speech-to-speech translation across 70 languages by preserving natural intonation and eliminating pauses 📚 Rolled out study notebooks to the @GeminiApp to help users reach their learning goals by analyzing uploaded class notes and quiz results to create personalized lessons and track progress 🌪️ Shared our updated weather models that use AI to help communities stay safe by predicting climate crises, like river floods, wildfires and cyclones

Available now for Google AI Pro and Ultra subscribers on mobile and web.

You can now create 60-second vertical Video Overviews in @NotebookLM. 🎬 Just upload your notes, study guides or links to get a short vertical video breaking down your most complex concepts. It’s powered by Nano Banana 2 Lite — our newest Gemini Image model.

gemini omni flash is here: our high-quality, cost-efficient model for video generation and conversational editing designed to support multimodal workflows, it enables you to refine videos using natural language and simple prompting start building with it today via ai studio and the gemini api

Nano Banana 2 Lite delivers text-to-image outputs in just 4 seconds. It’s designed for quicker ideation and workflows where speed and cost are the primary roadblocks.

Both models are available today in @GoogleAIStudio, the Gemini API, and the Gemini Enterprise Agent Platform. You can also find Nano Banana 2 Lite rolling out across AI Mode in Search, @GeminiApp, @NotebookLM, @GooglePhotos, @FlowByGoogle, @StitchByGoogle, and more. Learn more → goo.gle/3T9CzbG

Nano Banana 2 Lite 🤝 Gemini Omni Flash Now you can use the Interactions API to build with both models. We created this e-commerce applet in @GoogleAIStudio that turns static product shots into cinematic videos. Here’s how it works: 1⃣ Upload a product photo and set the vibe 2⃣ Nano Banana 2 Lite generates new, brand-accurate assets 3⃣ Gemini Omni Flash renders those images into a high-quality video

🌐 Gemini Omni Flash is our natively multimodal, high-quality model for video generation and conversational editing. It delivers: 🎬 Ability to use multimodal references 🗣️ Conversational video editing with natural language 🧠 World knowledge with real-world physics and logic ✍️ Synchronization for on-screen text and graphics 💸 $0.10 per second of video output (matching Veo 3.1 Fast) Now in public preview in: @GoogleAIStudio, the Gemini API, and Gemini Enterprise Agent Platform.

🍌 Nano Banana 2 Lite (gemini-3.1-flash-lite-image) is our fastest, most cost-efficient image model in the Nano Banana family. It delivers: ⚡ Images in 4 seconds (5x faster than Nano Banana 2) 💸 High-efficiency scaling at $0.034 per 1K images ✅ Reliable prompt adherence and legible text rendering Now GA in: @GoogleAIStudio and the Gemini API, and Gemini Enterprise Agent Platform. Rolling out across: AI Mode in Google Search, @GeminiApp, @NotebookLM, @GooglePhotos, @FlowByGoogle, @StitchByGoogle and more.

More users in the U.S. can now create personalized images for free. By choosing to connect your Google apps to Gemini in settings, Gemini is able to better understand your preferences, so it can automatically fill in the blanks and ground image creation in the things you care about most. You can choose to connect or disconnect your Google apps at any time, so you’re in control. Try it out today and share what you create in the replies 👇

For the full Q&A with @rseroter on our full-stack approach ↓ goo.gle/442E0uY

What are some of the benefits to working with full-stack AI? @rseroter: Because Google manages the entire stack — from running the underlying infrastructure all the way up to delivering @Gmail — there's massive system reliability. If a technical failure happens at one layer, our ownership of the platform allows us to catch it and handle it at another layer easily, rather than waiting for an external provider to fix it. There's also an economic advantage. Since we aren't paying third-party vendors for anything, customers don't have to absorb those fees, which means we can offer remarkably competitive pricing.

When Google first started working on AI, did we know we wanted a full-stack approach? @rseroter: It was absolutely a deliberate, decades-long strategy. Our bet on custom TPUs is already over 10 years old. We recognized early on that there’s massive value in owning our own supply chain and raw infrastructure when serving up the world's most important internet services. Owning that thread throughout the entire stack lets us deliver a level of service, performance and reliability that's very hard to achieve if you're at the mercy of multiple parties.

🧵Our unique full-stack approach to AI lets us deliver powerful, cost-efficient products to developers and everyday users. But what exactly does it mean when a technology system is "full-stack,” and why is it so important to our approach? We talked to @rseroter, senior director and chief evangelist at @googlecloud, to break it all down 👇 Where does the phrase “full-stack” come from, and what does it mean when we’re talking about tech? @rseroter: The term "full-stack" originally came out in software development a decade or so ago — usually in regard to applications. Historically, building an app required multiple specialized teams: a front-end developer to build beautiful user interfaces, a back-end developer to handle server-side logic and a dedicated database team. The concept of a "full-stack engineer" emerged to describe a developer who could work across all of these functions independently. Instead of constantly handing off components from one person to another, a full-stack engineer could take an idea from a rough concept all the way to a fully running piece of software.

🧵Up, up and away 🚀 Gemma 4, our most intelligent open models to date, just hit 200 million downloads. Since its launch 2.5 months ago, developers, researchers and startups have used Gemma 4 to fine-tune and deploy models across a variety of environments — from wearable robotic arms for physical assistance to enterprise-grade AI security.