100 things we announced at I/O 2026
Key Points
- gemini 3.5 flash
- gemini omni (video-first multimodal)
- antigravity generative ui & search agents
Summary
At I/O 2026 Google announced major updates across models, agents, search and developer tooling focused on multimodal generation, agentic workflows and deeper Search integration. Key launches include Gemini 3.5 Flash (low-latency, frontier intelligence), Gemini Omni (video-first multimodal generation with SynthID watermarking), Google Antigravity (agent-first platform and generative UI), and new Search agents/AI Search experiences. Several features and APIs are rolling out this summer to different Google AI subscription tiers and in consumer products (Gemini app, YouTube, Search). Engineers should prioritize the Gemini API, Antigravity primitives, and Search/Universal Cart integration points for early integration and experimentation.
Key Points
- Gemini 3.5 Flash: generally available via the Gemini API in Google AI Studio and Android Studio; designed for long‑horizon agentic tasks with Frontier-level quality at Flash latency. Reported metrics: Terminal-Bench 2.1 76.2%, GDPval‑AA 1656 Elo, MCP Atlas 83.6%.
- Gemini 3.5 Pro: in internal use now; wider rollout expected next month.
- Gemini Omni: new multimodal model starting with video outputs, supports references (image/text/video; audio voice references first), includes imperceptible SynthID watermark and verification in Gemini app/Chrome/Search; Omni Flash rolling out to Google AI Plus/Pro/Ultra and available in YouTube Shorts Remix and YouTube Create (18+).
- Platforms & APIs: Gemini API available in Google AI Studio and Android Studio; Antigravity is the agent-first platform powering agentic workflows and generative UI assembly.
- Search & agents: AI Mode now defaults to Gemini 3.5 Flash; new AI Search box supports text/images/files/videos/Chrome tabs and unifies AI Overviews + AI Mode. Information agents (24/7 monitoring and synthesized updates) and generative UI/mini‑apps via Antigravity will roll out this summer (subscriber-first for some features).
- Universal Cart & UCP: cross-product cart powered by Gemini models (price history, incompatibility checks, proactive suggestions) with Universal Commerce Protocol to enable streamlined checkout via Google Pay or retailer handoff.
- Gemini Spark & Daily Brief: Spark is a background personal agent (early testers → AI Ultra beta in U.S.); Daily Brief (personalized morning digest) rolling out to Google AI subscribers (18+) in the Gemini app.
- Practical integration notes: expect SDKs/APIs for Gemini models and Antigravity patterns; prioritize building around multimodal inputs/outputs and SynthID verification; anticipate subscription-gated features and staged regional rollouts.
- Privacy & safety: Personal Intelligence remains opt‑in for app connections; Spark and agents are designed to check before major actions; SynthID provides content provenance verification.