MOUNTAIN VIEW, CA — Google today unleashed a wave of new AI capabilities at its annual I/O conference, headlined by the release of Gemini 3.5 Flash and the launch of Gemini Omni, a model that can generate any output from any input. The company also gave a first look at Android XR glasses and detailed updates to AI Search, signaling an aggressive push to embed artificial intelligence across all its products.
“We are moving from AI that understands to AI that acts,” said Google CEO Sundar Pichai during the keynote. “Gemini 3.5 Flash combines frontier intelligence with speed, making it comparable to the best models but faster.”
Gemini Takes Center Stage
Gemini Omni
Demis Hassabis, CEO of Google DeepMind, unveiled Gemini Omni, a new model that can “create anything from any input.” It excels at simulating gravity and kinetic energy and combines Gemini intelligence with generative models like Nano Banana and Veo. Omni supports conversational language video editing—users can upload a video and edit any element by voice. “Omni starts with video, but eventually it will create any output from any input,” Hassabis said. The first Omni model, Gemini Omni Flash, is available today in the Gemini app.
Gemini 3.5 Flash and Pro
Gemini 3.5 Flash is available now across Google products and APIs. Pichai noted it “outperforms 3.1 Pro on nearly all benchmarks” while being faster. Internal testing of Gemini 3.5 Pro is underway, with a public release next month.
Redesigned Gemini App and AI Agents
The Gemini app sports a new Neural Expressive design language with fluid animations, haptic feedback, and custom regional dialects (coming in months). Paid subscribers get access to Gemini Omni for video creation. AI agents arrive, including a Daily Brief agent that curates a personalized digest. “These agents make Gemini proactive, not just reactive,” a Google spokesperson said. A Mac app now lets users select files in Finder and press the Function key to voice-command Gemini to act on them—for example, sending a dog’s information to a kennel via Gmail. Voice support and Gemini Spark arrive this summer.
Gemini for Science
Google introduced Gemini for Science, a suite of scientific tools, and the Co-Scientist collaborative AI research partner.
AI Content Identification
To combat deepfakes, Google is bringing C2PA content credentials to Gemini and Chrome. Tools can detect if an image was captured with a camera, generated by AI, or edited with AI. Users in Chrome can right-click on an image to see its provenance. “Transparency is critical as AI-generated content becomes indistinguishable from reality,” noted a Google product manager.
AI Search Updates
Google Search received AI enhancements: the “AI Overviews” feature now includes multi-step reasoning and real-time data integration. Shopping results become more personalized with generative summaries. “Search is becoming a conversational assistant,” Pichai said.
Android XR Glasses
Google previewed Android XR glasses, a partnership with Qualcomm and Samsung. The lightweight wearables overlay digital information onto the real world, controlled by voice and hand gestures. Early demos showed navigation arrows appearing on streets and real-time translations of signs. “This is the next computing platform,” said Sameer Samat, VP of Android. A developer beta launches today, with consumer release expected in 2027.
Background
Google I/O is the company’s annual developer conference, akin to Apple’s WWDC. The event offers a glimpse into Google’s product roadmap and how it will compete with rivals like Apple, Microsoft, and OpenAI. This year’s focus on Gemini and XR underscores a bet on ambient AI—intelligence that works seamlessly across devices and contexts.
What This Means
The announcements signal Google’s intent to own the AI stack from model to application. Gemini Omni could redefine content creation, while AI Search and Android XR extend Google’s reach into daily life. For users, expect more personalized, proactive assistance across Google’s ecosystem. For developers, new APIs and agents open opportunities for building on Gemini. “Google is no longer just an information company; it’s becoming an action company,” said analyst Maribel Lopez. “The next 12 months will determine if these bets pay off.”