Google supercharges Gemini with free Live visuals, Veo 3, and Imagen 4

Google is turning Gemini from a chatbot you text into a proactive agent that can see your world and run errands for you.

At Google I/O on May 20, the search giant announced that Gemini Live with real-time camera and screen sharing is now completely free for all users on Android and iOS. Instead of typing out queries, you can point your phone's lens at broken appliances or code on your monitor and talk it through. Google noted that conversational sessions with Gemini Live run five times longer than traditional text chats.

That multimodal push comes alongside a massive spec bump for creative and research workloads.

Talking videos and canvas vibe coding

Google rolled out Imagen 4 with sharper typography and up to 2K resolution, but the real showstopper is Veo 3. Google claims Veo 3 is the first video model with native support for background noise, sound effects, and character dialogue synchronized right out of the box. In early testing, The Verge called it wildly sophisticated yet an absolute "AI slop machine," noting that the model even generated unprompted character dialogue and eerie, clean street scenes.

Deep Research is also receiving a critical upgrade: users can now upload their own private PDFs and images to cross-reference with public web data, with future hooks into Google Drive and Gmail planned. Inside Canvas, users can generate interactive quizzes, export podcast-style Audio Overviews across 45 languages, or lean on Gemini 2.5 Pro for "vibe coding" working apps from natural language prompts.

Meanwhile, desktop users on Chrome will start seeing Gemini roll out to parse active webpages, with autonomous multi-tab browsing arriving down the road.

A $250 VIP pass for power users

Google also restructured its AI subscription tiers. It rebranded Gemini Advanced into Google AI Pro for $19.99 a month, bundling tools like NotebookLM and Flow, while offering college students in select countries a full school year for free.

Then came the heavyweight tier: Google AI Ultra. Priced at a staggering $249.99 per month (with a 50 percent discount for the first three months), it gives high-demand users the highest rate limits, access to Veo 3, the upcoming 2.5 Pro Deep Think mode, and early access to an experimental Agent Mode capable of carrying out complex, multi-step actions across Google apps with minimal human input.

Why it matters

For everyday users, getting free camera and screen sharing on mobile right now is an absolute win—troubleshooting physical gear or walking through complex screen workflows just got dramatically easier without spending a dime. On the other end of the spectrum, the $250-a-month Ultra tier is obviously not for casual tinkerers; it is an expensive sandbox aimed strictly at creative pros and enterprise pioneers.

For the industry, Google is flexing its unmatched distribution and ecosystem lock-in. By tying agentic workflows directly into Chrome, Maps, Calendar, and Workspace, Google is building defensive moats that standalone model makers simply cannot match.

We think launching native audio-video generation with Veo 3 and making visual voice chat free are remarkably sharp moves. However, Google's chaotic branding—juggling AI Mode, Deep Search, Deep Think, and Agent Mode—remains baffling. If it wants regular users to embrace agentic workflows, it will eventually need to clean up its own naming playbook.