Google has halved the entry price of its top-tier AI subscription, launched a cloud-resident personal agent and unveiled a new multimodal generation model capable of producing and editing video from any combination of inputs.
The announcements landed at Google I/O 2026 - the company's annual developer conference at Shoreline Amphitheatre in Mountain View - 10 years after Alphabet committed its product strategy to AI.
The moves represent the most concentrated product push in the company's AI history, spanning subscriptions, search architecture, video generation and ambient computing.
Google AI Ultra now starts at US$100 per month - down from $250 - and is pitched at developers, technical leads, knowledge workers and advanced creators, while the previous top-tier price drops to $200 with identical capabilities.
Its Gemini app now crosses 900 million active users - a 2x growth over the past year, processing 9.7 trillion tokens per month.
Bright Spark
The entry-level $100 plan is not a stripped-back offering.
It bundles a 5x higher usage limit in the Gemini app than the existing AI Pro tier, 20 terabytes of cloud storage and a YouTube Premium individual plan alongside beta access to Gemini Spark - the announcement that drew the most scrutiny from developers in attendance.
Spark is a cloud-resident AI agent that tracks Gmail, Calendar, Google Tasks and connected third-party applications in the background without requiring the Gemini app to stay open, running on the company's Gemini 3.5 Flash model and Antigravity 2.0 development platform.
The agent handles multi-step workflows autonomously - drafting correspondence, assembling morning briefings, populating spreadsheets from voice instructions - and because it runs server-side, it keeps executing long after a device is closed or locked.
"Gemini Spark is a 24/7 AI agent that helps you navigate your digital life, takes action on your behalf and is under your direction," Google VP Shimrit Ben-Yair said.
"Gemini Spark can connect the dots across your Google products and take complex tasks off your plate so you can be more present for what matters.
Spark will roll out to trusted testers this week, and we're planning to roll it out as a Beta for Google AI Ultra subscribers in the U.S. next week."
On the model side, Mountain View introduced Gemini Omni - a new family that marries Gemini's reasoning depth with generative media output, ingesting images, audio, video and text to produce high-quality footage anchored in real-world knowledge.
The first release in the family - Gemini Omni Flash - supports conversational editing across multiple turns, meaning creators can iteratively refine a scene, shift camera angles or swap out specific objects without rebuilding from scratch.
All footage produced with Omni carries Google's SynthID digital watermark, verifiable through the Gemini app, Chrome and Search.
Search itself absorbed its most significant architectural rework in years, with AI Mode now powered by Gemini 3.5 Flash globally, a redesigned query box that expands as entries lengthen and background Search Agents that keep tabs on topics continuously for subscribers on paid tiers.
