At the annual Google I/O 2026 event today, Google introduced a range of new AI tools, including Gemini 3.5 Flash, a new AI model, and Antigravity 2.0, an advanced AI coding platform. The company showcased how artificial intelligence is becoming a bigger part of software development and automation. The announcement came during a presentation by Varun Mohan, who described Antigravity 2.0 as an “unabashedly agent-first” platform focused on helping developers build software with teams of autonomous AI agents. During the live demo, Google revealed that it had tested the platform by assigning it a massive challenge: create an entirely new operating system from scratch.
One of the biggest highlights was Google’s claim that Antigravity 2.0 managed to build the core framework of a working operating system in just 12 hours using multiple AI agents working together.
Antigravity 2.0 builds operating system
During a live demonstration, Google revealed that Antigravity 2.0 used 93 AI sub-agents to create the operating system. According to the company, the system generated 2.6 billion tokens while completing the project at a cost of under $1,000 in AI processing expenses. The demo became more interesting when Google attempted to run the classic game Doom on the AI-created operating system. Initially, the game failed because keyboard drivers were missing. Google then asked Antigravity to generate the missing drivers live on stage, after which the game reportedly became playable.
Google said Antigravity 2.0 is designed as an “agent-first” platform where multiple AI agents can independently handle different tasks at the same time. The company also confirmed that the platform will now be available as a standalone desktop application with command-line support for developers.
Gemini 3.5 Flash launched
Alongside Antigravity, Google also launched Gemini 3.5 Flash, the latest version of its AI model. CEO Sundar Pichai said the new model is faster and more capable than previous versions. Google claims Gemini 3.5 Flash can generate responses nearly four times faster than competing AI systems while operating at lower costs. The company said the model is specially designed for advanced AI agents, coding, and longer workflows.
The company also announced that Gemini 3.5 Flash will become the default Gemini model starting immediately, while a more advanced Gemini 3.5 Pro version is expected to launch later.
Google doubles down on AI future
This year’s Google I/O event showed the company’s growing focus on AI-powered tools across products and services. From coding assistants to autonomous AI agents, Google is positioning Gemini and Antigravity as key technologies for the future of software development.