I/O 2025: Gemini App and Generative Media

Year: 2025 · ▶ Watch on YouTube

Josh Woodward (Presenter) · Jason Baldridge (Presenter)

Switch language → zh

Segments (8)

  • 00:00:18 · The Vision for a Universal AI Assistant — Josh Woodward
    • Introducing the goal to make Gemini the most personal, proactive, and powerful AI assistant, built on the principles of being personal, proactive, and powerful.
  • 00:03:45 · Gemini Live and New Capabilities — Josh Woodward
    • Announcing that Gemini Live is becoming free on Android and iOS with new camera and screen sharing capabilities, plus integrations with other Google apps.
  • 00:05:20 · Deep Research, Canvas, and Gemini in Chrome — Josh Woodward
    • Detailing new features for deep research with file uploads, co-creation in Canvas, and a new Gemini integration directly within the Chrome browser.
  • 00:07:44 · Imagen 4 and Veo 3: The Next Generation of Creative Models — Josh Woodward
    • Unveiling Imagen 4 for advanced image generation and Veo 3, a state-of-the-art model for generating high-quality video with native audio and dialogue.
  • 13:02:48 · Generative Media and Creative Collaboration — Jason Baldridge
    • Discussing how generative media is expanding creativity through tools like Music AI Sandbox and collaborations with filmmakers like Darren Aronofsky.
  • 15:56:56 · SynthID: Watermarking and AI Safety — Jason Baldridge
    • Highlighting Google’s commitment to AI safety by expanding SynthID for watermarking generative content and providing a new detector tool.
  • 20:14:26 · Flow: A New AI Filmmaking Tool — Josh Woodward
    • Introducing and demonstrating ‘Flow,’ a new tool for creatives that combines Veo, Imagen, and Gemini to build stories and films from prompts and ingredients.
  • 25:06:45 · New Google AI Subscription Plans — Josh Woodward
    • Announcing the new Google AI Pro and Google AI Ultra subscription plans, which offer tiered access to the latest and most powerful AI features.

Products Announced (12)

  • 00:00:31 · Gemini (Updated Vision)
    • Personal · Proactive · Powerful
    • Core experience is part of Google’s ecosystem.
  • 00:03:48 · Gemini Live (Free on Android & iOS)
    • Natural, interactive voice conversations · Camera and screen sharing · Integrations with Calendar, Maps, Keep, Tasks
    • Rolling out free of charge starting today.
  • 00:05:27 · Deep Research (Updated Feature)
    • Upload your own files to guide research · Research across Google Drive and Gmail
    • File upload available today; Drive/Gmail integration coming soon.
  • 00:06:01 · Canvas (Updated Feature)
    • Interactive co-creation space · Transform reports into webpages, infographics, quizzes · Vibe code interactive applications
    • Available today.
  • 00:06:58 · Gemini in Chrome (New Integration)
    • Personal AI assistant for browsing · Understands context of the current webpage · Summarize and compare information on a page
    • Rolling out this week to Gemini subscribers in the US.
  • 00:07:53 · Imagen 4 (New Model)
    • High-quality image generation · Improved text and typography rendering · Creative font and layout choices
    • Available in the Gemini app starting today.
  • 00:09:37 · Veo 3 (New Model)
    • State-of-the-art video generation · Native audio generation (sound effects, dialogue) · Photorealistic quality and physics understanding
    • Available today in the Google AI Ultra plan.
  • 13:34:52 · Music AI Sandbox (Professional Tool)
    • Powered by Lyria 2 model · Explore generative music possibilities · Create instrumental beds and song ideas
    • Available to musicians and creators.
  • 15:59:02 · SynthID (Expanded)
    • Embeds invisible watermarks in generative media · New detector tool to identify watermarks · Works on images, audio, text, and video
    • Detector rolling out to early testers today.
  • 20:47:58 · Flow (New Tool)
    • AI filmmaking tool for creatives · Combines Veo, Imagen, and Gemini · Builds stories from prompts, images, and text
    • Launching today.
  • 25:12:47 · Google AI Pro (New Subscription Plan)
    • Access to Gemini 2.5 Pro & Veo 2 · Gemini in Gmail, Docs, etc. · 2 TB Storage
    • $19.99 / month, available globally.
  • 25:35:36 · Google AI Ultra (New Subscription Plan)
    • Highest rate limits · Early access to new features like Flow with Veo 3 · Includes YouTube Premium and 30 TB Storage
    • $249.99 / month, available in the US today.

Commitments / Timelines (10)

  • 00:02:55 (Starting soon) — Ability to add more personal context from across Google to Gemini.
  • 00:04:20 (Rolling out today) — Gemini Live with camera and screen sharing will be free on Android and iOS.
  • 00:04:31 (In the coming weeks) — Connect Gemini Live to apps like Calendar, Maps, Keep, and Tasks.
  • 00:05:30 (Starting today) — Deep Research will allow uploading your own files.
  • 00:05:38 (Soon) — Deep Research will allow searching across Google Drive and Gmail.
  • 00:07:25 (This week) — Gemini in Chrome will roll out to Gemini subscribers in the US.
  • 00:07:45 (Starting today) — Imagen 4 image generation model is being brought into the Gemini app.
  • 00:09:50 (Today) — Veo 3 video generation model is available.
  • 20:49:03 (Today) — Flow AI filmmaking tool is launching.
  • 25:53:11 (Today) — Google AI Ultra plan is available in the US.

Demos (8)

  • 00:02:04 ✓ · Proactive Assistant for Student — Josh Woodward
    • A mockup of a phone screen showing Gemini proactively creating a personalized physics quiz based on an upcoming calendar event and user’s notes.
  • 00:04:39 ✓ · Gemini Live with Camera — Josh Woodward
    • A mockup showing a user pointing their phone camera at a handwritten shopping list, and Gemini Live deciphering it and adding the items to a Google Keep list.
  • 00:05:47 ✓ · Deep Research and Canvas — Josh Woodward
    • A user uploads a detailed report on comets, and Canvas transforms it into an interactive comet simulation application with a single click.
  • 00:07:09 ✓ · Gemini in Chrome — Josh Woodward
    • A user on a webpage for a campsite asks Gemini, ‘Which campsites have river access?’ and Gemini provides a summarized answer based on the page’s content.
  • 00:08:35 ✓ · Imagen 4 Poster Generation — Josh Woodward
    • A series of prompts are used to generate a detailed and stylized music festival poster featuring a robotic dinosaur DJ, with Imagen 4 correctly rendering text and making creative font choices.
  • 00:09:51 ✓ · Veo 3 Video and Audio Generation — Josh Woodward
    • Two videos generated by Veo 3: one of a wise owl and a nervous badger having a conversation, and another of an old sailor on a boat, both with generated dialogue and ambient sound.
  • 13:53:00 ✓ · Music AI Sandbox with Shankar Mahadevan — Shankar Mahadevan (in video)
    • Grammy-winning artist Shankar Mahadevan uses the Music AI Sandbox to generate an instrumental bed, which he then uses as inspiration to compose and record a new song.
  • 20:55:20 ✓ · Flow AI Filmmaking Tool — Josh Woodward
    • A demonstration of the ‘Flow’ tool, where images of an old man and a car are used as ‘ingredients’ to generate a short film about the man building a flying car with help from a giant chicken.

Notable Quotes (8)

  • 00:00:41 — Josh Woodward:

    Our goal is to make Gemini the most personal, proactive, and powerful AI assistant.

  • 00:02:25 — Josh Woodward:

    That’s not just helpful, it’s going to feel like magic.

  • 00:06:49 — Josh Woodward:

    This is the power to transform anything.

  • 11:22:37 — Josh Woodward:

    We’re entering a new era of creation with combined audio and video generation that’s incredibly realistic.

  • 13:15:10 — Jason Baldridge:

    Whether you’re a creator, a musician, or a filmmaker, generative media is expanding the boundaries of creativity.

  • 17:32:57 — Darren Aronofsky:

    I don’t think that ever changes.

  • 24:30:20 — AI Filmmaker:

    I’m not forcing it, I’m just finding it. And that’s when I know I’m in the right place.

  • 26:00:00 — Josh Woodward:

    You can think of this Ultra plan as your VIP pass for Google AI.

Visual Signals (Beyond the Transcript)

On-Screen Text Moments (18)

  • 00:00:03 · Google I/O logo
    • Brands the event.
  • 00:00:13 · Introducing Josh Woodward
    • Identifies the first speaker.
  • 00:00:31 · Gemini logo
    • Introduces the central product of the presentation.
  • 00:00:45 · Personal Proactive Powerful
    • Outlines the three core principles of the Gemini assistant vision.
  • 00:01:13 · Personal context
    • Names the feature that allows Gemini to use user’s personal data.
  • 00:03:03 · Gemini 2.5 Pro
    • Names the underlying model powering the new features.
  • 00:03:49 · Gemini Live
    • Announces the Gemini Live feature set.
  • 00:05:28 · Deep Research
    • Introduces the research feature.
  • 00:06:02 · Canvas
    • Introduces the co-creation tool.
  • 00:07:01 · Gemini in Chrome
    • Announces the browser integration.
  • 00:07:54 · Imagen 4
    • Announces the new image generation model.
  • 00:09:38 · Veo 3
    • Announces the new video generation model.
  • 13:45:34 · Music AI Sandbox
    • Names the tool for professional musicians.
  • 16:33:51 · labs.google/synthid
    • Provides a URL for users to sign up for the SynthID detector.
  • 20:49:03 · Flow
    • Announces the new AI filmmaking tool.
  • 24:45:48 · flow.google
    • Provides the URL for the new Flow tool.
  • 25:13:28 · Google AI Pro plan details and pricing ($19.99/month)
    • Details the features and cost of the new Pro subscription tier.
  • 25:37:31 · Google AI Ultra plan details and pricing ($249.99/month)
    • Details the features and cost of the new Ultra subscription tier.

Stage Moments (8)

  • 00:00:07 · Josh Woodward walks onto the stage to applause from a large, live audience in an outdoor amphitheater.
  • 00:04:25 · Audience applauds the announcement that Gemini Live is rolling out for free.
  • 00:07:31 · Audience applauds the announcement of Gemini in Chrome.
  • 00:07:56 · Audience applauds the announcement of Imagen 4.
  • 00:09:56 · Audience applauds enthusiastically for the announcement of Veo 3.
  • 13:07:23 · Jason Baldridge walks onto the stage.
  • 20:15:20 · Josh Woodward walks back onto the stage, joining Jason Baldridge.
  • 22:51:30 · Audience applauds and cheers for the ‘flying chicken car’ short film created with Flow.

Visual Demos (8)

  • 00:02:04 · A UI mockup of Gemini’s proactive capabilities.
    • A phone lock screen shows a notification from Gemini about an upcoming physics exam, offering to start a practice quiz on thermodynamics.
  • 00:08:03 · Images generated by Imagen 4.
    • A series of high-quality, stylized images are shown, including a woman in a flowing green dress, a papercraft bird, a close-up of a dandelion with water droplets, and skiers on an ice cream cone.
  • 00:08:37 · A poster generated by Imagen 4.
    • A vibrant, psychedelic poster for ‘THE DINO MUSIC FESTIVAL 2025’ featuring a robotic T-Rex DJ. The text is rendered clearly and creatively, with ‘DINO’ made of what looks like bones.
  • 10:47:20 · A video generated by Veo 3 with audio.
    • A short animated scene of a wise owl and a nervous badger in a mystical forest. Both characters speak with generated voices, and there are ambient forest sounds.
  • 11:31:30 · A photorealistic video generated by Veo 3 with audio.
    • A cinematic shot of an old sailor on a boat, looking out at the ocean. He speaks with a generated, gravelly voice, and the sound of the ocean is present.
  • 17:19:20 · A trailer for the film ‘Ancestra’.
    • A professionally produced film trailer combining live-action hospital scenes with surreal, Veo-generated visuals of cellular processes, black holes, and abstract imagery, telling the story of a mother and her newborn child.
  • 20:55:20 · The UI of the ‘Flow’ AI filmmaking tool.
    • A storyboard-like interface where a user provides images (an old man, a car) and text prompts to generate a sequence of video clips, including creating a 10-foot-tall chicken in the back seat and making the car fly.
  • 23:12:00 · A short film created with ‘Flow’.
    • A montage of surreal and creative short clips from various AI filmmakers, demonstrating the capabilities of the Flow tool, including a woman with a lava lamp backpack and surgeons operating in the back of a taxi.

Production Signals (5)

  • 00:00:00 · High-energy, animated intro sequence for Google I/O.
  • 00:00:07 · Live presentation on a large, custom-built outdoor stage in front of a large audience.
  • 13:53:00 · Cut to a pre-recorded, professionally shot video segment featuring musician Shankar Mahadevan in his studio.
  • 17:19:20 · Cut to a pre-recorded segment featuring filmmaker Darren Aronofsky, followed by a cinematic trailer for the film ‘Ancestra’.
  • 23:12:00 · A fast-paced, pre-recorded montage showcasing the creative output of several AI filmmakers using the new tools.

Key Topics

Generative AI · AI Assistants · Gemini · Multimodality · Video Generation · Image Generation · Audio Generation · Creative Tools · AI Filmmaking · Personalization · AI Safety · Developer Tools · Google Chrome · Android · Subscription Models

Takeaways

  • Google is positioning Gemini as a universal, proactive, and deeply personal AI assistant that will be integrated across its entire product ecosystem, moving beyond simple reactive commands.
  • A major focus is on empowering creativity with a suite of new and updated generative models, including Imagen 4 for images and the highly advanced Veo 3 for video with integrated audio generation.
  • Google is launching ‘Flow,’ a new AI filmmaking tool, and collaborating directly with professional creators like Darren Aronofsky to build tools that meet the needs of the creative industry.
  • The Gemini experience is being stratified into new subscription tiers, ‘Google AI Pro’ and ‘Google AI Ultra,’ offering different levels of access, rate limits, and early features to monetize its most advanced capabilities.
  • Multimodality is central to the new Gemini, with features like Gemini Live using voice and camera input simultaneously, and models like Veo 3 combining video, sound effects, and dialogue generation.
  • AI safety remains a stated priority, with the expansion of the SynthID watermarking technology to more content types and the release of a detector tool to promote transparency.
  • The line between different generative tools is blurring, with products like Flow and Canvas combining image, video, and text generation into a single, cohesive creative workflow.