Google

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.

For a list of the key moments that got us here ⬇️ goo.gle/3SRSATN

📍2018: Lens in Google Search Google Lens makes everyday life searchable, letting you use your smartphone camera to ask about the world around you. With Lens, you can identify objects, translate text and pull up product links in real time without typing a query into the search box.

This week marks 25 years since we launched Google Images 🎉 To celebrate this milestone, we’re taking a look back at our major visual innovations that took Search from a text-based experience to one that allows you to visually explore the world. Plus, we’re introducing two new ways to explore and create visual content in Search ↓🧵

Celebrating 25 years since we launched Google Images with even more visual search innovation! 🎂 + Introducing a new home for Google Images on desktop: immersive, dynamic & tailored to your interests + Bringing Nano Banana 🍌 image generation into Search so you can ask to generate an image and get it right on the Search results page

📣 New Google Calendar label colors have entered the chat!

@googlecalendar Tag yourself we're cobalt

For a full breakdown on the new features ⬇️ goo.gle/3SRSATN

We’re introducing two new ways you can explore and create visual content on Search 👇 1️⃣ Discover and organize visuals in Google Images on desktop When you open the brand-new browseable home for Google Images, you’ll now see a dynamic, immersive image gallery that’s updated in real time and intelligently tailored to your unique interests. You’ll be able to browse and save ideas to your collections, and they’ll appear as tabs above the main gallery, making it easy to jump back in and continue searching. This new experience will roll out on desktop over the coming weeks in the U.S. in English. Just make sure you’re signed into your Google Account to try it out. 2️⃣ Create images in AI Overviews in Search Using our latest Nano Banana model, you can turn any text prompt into a custom visual — directly in the Search results page. Simply go to google.com or open the Google app, type in the image you want to create and that's it. Image generation in AI Overviews will roll out over the coming weeks in English, for all the regions that currently support image creation in AI Mode.

@DiscussingFilm the best search results come with a surprise 🐎

‘THE ODYSSEY’ has an interactive Trojan Horse when you search for the film on Google.

@NolanAnalyst everyone shhhhh

Google has launched a special interactive animation when you search for “The Odyssey.”

@DanielSmidstrup @googlegemma @huggingface Lightning fast ⚡

@Google @googlegemma @huggingface 5x faster in 6 days is wild, humans directing and agents grinding looks like a serious combo :D

100+ AI agents and developers just teamed up to make inference 5x faster on our Gemma 4 open models ⚡ Our @GoogleGemma team partnered with @huggingface for a 6-day sprint to see how far we could push inference speed. Agents pooled resources and self-policed against "lazy" shortcuts, while humans stayed in the loop to provide direction and research taste. See how they did it ↓

@ComfyUI Telekinetic powers are just a prompt away

Gemini Omni Flash can manipulate objects and environments in existing videos using simple text prompts. Sometimes it takes a bit of iteration and specific prompting, but it opens up a lot of creative possibilities. To try this workflow, link below 👇

@FOXSports Instructions clear 🥁

When Erling Haaland says to Google his name, you Google his name.

A model’s chain of thought acts like a scratch pad, offering a window into its reasoning. 📝 On the latest episode of our podcast, host @fryrsquared sits down with @NeelNanda5 to explore interpretability – the science of reverse engineering how neural networks learn and think. Timecodes: 00:00 Introduction 02:41 Motivation for interpretability research 04:01 Mechanistic interpretability 08:14 Chain of thought monitoring 18:14 Interpretability techniques 35:00 Auditing models for safety 48:53 What comes next for interpretability

@JustSha77933233 Not as well as they do ⚽

@Google Well played Google. Well played

Rowing straight to the search bar 🌊