Google used the keynote of its I/O developer conference on May 19 to ship a coordinated three-product Gemini update — Gemini 3.5 Flash, the new Omni model, and the Spark general-purpose agent. The release is, in scope, the most expansive single Gemini moment the lab has staged.
Gemini 3.5 Flash is the lab’s new lightweight model in the family, positioned as cutting-edge capabilities at roughly half — and, on some workloads, close to one-third — the price of comparable frontier models. The Flash variant is the consumer-facing successor to the 3.1 Flash-Lite generally-available release that landed earlier in May.
Gemini Omni is the more architecturally ambitious of the three. Google described Omni as a leap forward in world understanding, multimodality, and editing — a model that lets users generate any output from any input, starting with video. The framing reads as an explicit response to OpenAI’s GPT-5.5 Instant launch two weeks earlier, which focused on reliability inside the default text-and-voice surface rather than expanding the modality envelope.
Gemini Spark is the agent. Google described it as a new general-purpose AI agent that reasons across information in connected apps. It launched in beta, available first to trusted testers and Google AI Ultra subscribers. The Spark positioning slots Google into the agent-first messaging that Microsoft (Agent 365), Anthropic (Managed Agents), and Salesforce (Agentforce) have all foregrounded over the past two weeks.
The three-product release is consistent with the lab’s recent cadence — a model, a modality push, and an agent surface, shipped together. For the broader competitive picture, the I/O drop signals that Google intends to compete on all three frontiers simultaneously rather than picking one.