Google has introduced Gemini 3.5 Flash, touted as the “most powerful AI model in programming and agent creation.
Meet Gemini 3.5 Flash — our strongest agentic and coding model yet.
It delivers frontier-level performance at 4x the speed of comparable frontier models — often at less than half the cost.
Generally available, starting today. 🧵#GoogleIO pic.twitter.com/jLhqozutwG
— Google (@Google) May 19, 2026
It can independently code, manage research projects, and create an operating system from scratch.
Gemini 3.5 Flash is the first LLM in the new Gemini 3.5 lineup—a family of models that combines high capability with action execution. The neural network is designed for complex multi-step tasks and long workflows.
Google claims that 3.5 Flash generates four times more tokens per second compared to other advanced models. The Gemini 3.5 Pro version is promised for release next month.
The company presents the launch as a shift from conversational AI to agentic AI. The model is designed for tasks where the system needs not only to answer questions but also to execute commands sequentially.
For certain actions like sending emails or financial transactions, the agent will request additional permission.
Gemini 3.5 Flash is being integrated into the Gemini app and other company products. In a developer blog, Google writes that the model should “accelerate the transition from prompts to actions.”
The corporation also introduced Gemini Spark—a personal AI assistant that helps with daily tasks. It operates on Gemini 3.5 and can freely interact with Google services.
Other Releases
Antigravity 2.0
Antigravity 2.0 is another new release. The application serves “as a central platform for interacting with agents.” It allows for the coordination of multiple digital assistants to perform tasks in parallel.
It supports dynamic sub-agents for different workflows, scheduled tasks for background automation, and integration with the ecosystem, including Google AI Studio, Android, and Firebase.
New features include:
- Antigravity CLI—provides a “lightweight and high-performance interface, allowing instant creation of new agents without using a graphical user interface;
- Antigravity SDK—programmatic access to the same set of agent tools used in Google products;
- Antigravity in Gemini Enterprise Agent Platform—simplifies corporate workloads, allowing Google Cloud customers to connect the service directly to their projects.
A new Google AI Ultra plan is introduced at $100 per month, offering a usage limit in Antigravity five times higher than Google AI Pro.
Omni
The company introduced Gemini Omni—a model designed to “turn fantasies into reality.”
“Through the seamless integration of text, images, and video materials, the neural network allows for the effortless creation of stunning high-quality videos,” the release states.
Users can apply cinematic zoom effects or change backgrounds using a text prompt.
Google Pics
Google Pics is a new tool for creating and editing images. It allows users to create “virtually anything—from party posters to infographics.”
Pics automatically highlights objects in photos and analyzes their relationships.
Managed Agents in Gemini API
Google introduced managed agents in the API Gemini. With a single call, a digital assistant can be launched that can think logically, use tools, and execute code in an isolated Linux environment.
AI assistants operate on the Antigravity platform, built on Gemini 3.5 Flash and accessible through API Interactions and in Google AI Studio.
Google also announced work on smart glasses and showcased the design of the upcoming fall collection.
We’re partnering with @Samsung, @_GentleMonster_ and @WarbyParker on new intelligent eyewear.
Here’s a sneak peek at two designs from this fall’s upcoming collections.#GoogleIO pic.twitter.com/f10T1qATxJ
— Google (@Google) May 19, 2026
Back in May, Google DeepMind introduced an AI-powered computer mouse cursor based on Gemini—the company aims to rethink a technology that has remained unchanged for over half a century.