At its annual I/O developer conference at the Shoreline Amphitheatre in Mountain View on May 19, 2026, Google announced a wide slate of AI model releases, agentic products, and hardware partnerships. CEO Sundar Pichai opened the keynote by reporting that <cite index="7-3">the Gemini app now serves more than 900 million people every month across 230 countries and more than 70 languages, up from 400 million at the same time last year, with daily queries growing sevenfold in that period</cite>. <cite index="5-10">Pichai revealed the company is now processing more than 3.2 quadrillion tokens per month, a massive increase from 480 trillion tokens per month at I/O 2025</cite>.
Gemini 3.5 Flash and Omni
The centerpiece model release is Gemini 3.5 Flash, which Google positions as a frontier-grade model at Flash-tier latency. <cite index="2-18">Gemini 3.5 Flash is generally available today via Google Antigravity, the Gemini API in Google AI Studio, and Android Studio</cite>. According to Google's published benchmarks, <cite index="2-21">it outperforms Gemini 3.1 Pro on coding and agentic benchmarks including Terminal-Bench 2.1 (76.2%), GDPval-AA (1656 Elo) and MCP Atlas (83.6%)</cite>. <cite index="5-9">The company says 3.5 Flash delivers four times the output token generation speed of competing frontier models</cite>, with <cite index="3-28">a 12x speed-up when running inside Antigravity 2.0</cite>. <cite index="2-41,2-42">Gemini 3.5 Pro is in internal use and is expected to roll out next month</cite>.
Google also introduced Gemini Omni, a family of natively multimodal generative models. <cite index="7-13,7-14,7-15">Omni is designed to accept any combination of text, images, video, and audio as input and produce any output in return, with video launching first. Google DeepMind's Koray Kavukcuoglu described it as a generalization of Veo, trained as multimodal from the ground up. All content it produces carries Google's imperceptible SynthID digital watermark</cite>.
Gemini Spark and Universal Cart
The agentic layer of the announcement centers on Gemini Spark, described by Google as a personal agent that acts on a user's behalf. <cite index="3-31,3-32,3-33">Spark runs on virtual machines through Google Cloud and is able to operate 24/7, with no need to keep a laptop open. It is accessible through the Gemini app, with options to email or message it, and uses Gemini 3.5 Flash and Antigravity to work on long-running tasks in the background</cite>.
The most direct competitive move against Amazon is Universal Cart. <cite index="1-23,1-24">Universal Cart is a Gemini-powered shopping cart and agentic hub for shopping in the Gemini app, YouTube, and Gmail. Once a user adds a product, Google finds deals, price drops, price history insights, and stock alerts</cite>. <cite index="8-5,8-6,8-7,8-8">It integrates with Google Wallet to account for payment method perks and loyalty information. Users can check out via Google Pay or transfer the cart to a merchant's site. Universal Cart will roll out to Search and the Gemini app this summer, with YouTube and Gmail coming soon after</cite>.
Hardware and pricing
On hardware, <cite index="8-9,8-10,8-11">Google announced new Android XR audio glasses for this fall that speak information privately into the ear and allow users to take photos and listen to music hands-free. Samsung, Warby Parker and Gentle Monster are the first hardware partners, and the audio glasses will connect with both Android and iOS devices</cite>.
Google also restructured its subscription tiers. <cite index="13-15,13-16,13-17">Google AI Ultra now starts at $100 per month—down from $250—aimed at developers, technical leads, knowledge workers, and advanced creators. Google simultaneously lowered its previous $250 top tier to $200. The $100 plan includes a 5x higher usage limit than the $20 AI Pro tier, 20 terabytes of cloud storage, YouTube Premium, and beta access to Gemini Spark starting next week for US subscribers</cite>.