Gemini 3.5 Flash: Frontier Intelligence, Free for All
Gemini 3.5 Flash launched as the first model in the Gemini 3.5 series, available immediately on 19 May 2026. It is optimised for agents and coding, delivering frontier intelligence without speed trade-offs.
The model is accessible at no cost in the Gemini app and AI Mode in Search. Paid access is available through Google Antigravity, the Gemini API in AI Studio, and Android Studio.
Benchmarks vs. Gemini 3.1 Pro
| Benchmark | Score |
|---|---|
| Terminal‑Bench 2.1 | 76.2% |
| GDPval‑AA | 1656 Elo |
| MCP Atlas | 83.6% |
| CharXiv Reasoning (multimodal) | 84.2% |
Output is 4× faster than other frontier models.
Agentic capabilities handle long‑horizon tasks. Demonstrations included migrating a legacy codebase to Next.js, synthesising a research paper into a playable game in six hours, and generating multiple UX checkout approaches in 60 seconds.
Enterprise partners already using it: Shopify runs parallel subagents for merchant growth forecasts, Macquarie pilots onboarding acceleration on 100+‑page documents, and Xero deploys autonomous multi‑week tax‑form workflows. Gemini 3.5 Pro is in development and will roll out in June 2026.
AI Search Reimagined: New Search Box and 1B‑User Milestone
AI Mode surpassed 1 billion monthly users and now runs on Gemini 3.5 Flash globally. The search box itself received its largest overhaul in over 25 years, now accepting text, images, files, video, and Chrome tabs, then reasoning across all inputs.
AI Overviews and AI Mode are unified into one seamless flow: a question leads to an AI Overview, then to follow‑up in AI Mode — all with source links. This live on desktop and mobile worldwide.
New information agents operate 24/7, monitoring blogs, news, social media, and fresh data on finance, shopping, and sports. They deliver synthesised updates and can take action. Multiple agents run simultaneously. These roll out to Google AI Pro and Ultra subscribers in summer 2026.
Generative UI, powered by Antigravity and Gemini 3.5 Flash, builds custom interactive layouts, tables, graphs, and simulations on the fly. Subscribers will later be able to create persistent mini apps for ongoing tasks like wedding planning or moving house.
Gemini Omni: Multimodal Video Generation from Any Input
Gemini Omni is a new model that combines Gemini intelligence with generative media capabilities. It can create video output from image, text, video, or audio references. Audio input is initially limited to voice.
The model integrates physics understanding — gravity, kinetic energy, fluid dynamics — with cultural and historical knowledge. All output includes a SynthID watermark, verifiable through the Gemini app, Gemini in Chrome, and Search.
Gemini Omni Flash rolls out to Google AI Plus, Pro, and Ultra subscribers, and is available at no cost for users 18+ inside YouTube Shorts Remix and the YouTube Create app. The full Gemini Omni model remains for subscribers in the Gemini app.
In‑app features include conversational editing (cinematic zooms, background changes, custom AI avatars), Google Flow for character‑consistent blended content, and YouTube Shorts Remix for one‑tap video transformations. Future updates target any output type from any input.
Antigravity 2.0: Agent‑First Platform for Parallel Workflows
Antigravity 2.0 is a standalone desktop application that serves as a central hub for orchestrating multiple agents in parallel. One agent can build a website while another generates brand assets.
Key additions:
- Antigravity CLI – a lightweight, terminal‑based surface for instant agent creation.
- Antigravity SDK – programmatic access to the same harness, with custom behaviour and self‑hosting options.
- Core primitives: subagents, hooks, and asynchronous task management.
- Native voice support for Gemini audio models.
Multi‑day engineering efforts are collapsed to hours through subagent teamwork. Existing Gemini CLI users are encouraged to migrate to the Antigravity CLI.
Enterprises can connect Antigravity directly to Google Cloud projects under enterprise terms. A roll‑out to existing Gemini Enterprise customers is planned in coming months.
Antigravity also powers generative UI in Search: it builds custom interactive layouts and simulations in real time, free for all users from summer 2026.
Universal Cart: Intelligent Shopping Across Google Surfaces
Universal Cart is an AI‑powered shopping hub that functions across Search, the Gemini app, YouTube, and Gmail. It improves automatically as Gemini models evolve.
Items can be added while browsing Search, chatting with Gemini, watching YouTube, or reading Gmail. The cart runs background actions: finding deals, price drops, and back‑in‑stock alerts. It proactively flags product incompatibilities and suggests alternatives.
Built on Google Wallet, the cart understands payment‑method perks, loyalty information, and merchant offers, then intelligently selects the best payment option. The Universal Commerce Protocol (UCP) enables two‑tap checkout for many brands, or transfers items to a retailer’s site seamlessly.
Roll‑out starts in Search and the Gemini app in summer 2026; YouTube and Gmail will follow later.
Gemini Spark: 24/7 Personal AI Agent
Gemini Spark is a persistent personal AI agent that runs in the background on a phone or laptop, even when the device is off. It acts on the user’s behalf under user direction, checking before major actions.
The agent runs on Gemini 3.5 and the Antigravity platform, operating autonomously but requiring user authorisation for critical decisions. A beta programme begins for Google AI Ultra subscribers in the U.S. the week of 19 May 2026, starting with trusted testers.
Upcoming capabilities include the ability to text or email Spark directly, create custom sub‑agents for specialised tasks, and authorise payments — within budget and merchant parameters set by the user.
