Google's Gemini 3.5 Pro, unveiled at Google I/O on May 19, is in limited preview for select Vertex AI enterprise customers and targeted for general availability in June 2026. The model aims for a 2-million-token context window, Deep Think reasoning, and frontier multimodal capability -- the high-end use cases the old Gemini Ultra tier used to cover.
As of mid-June, Pro had not yet shipped to the public Gemini app, AI Studio, or the consumer subscription, and Sundar Pichai's 'wait another month' framing at I/O reportedly drew groans from a crowd hoping for immediate access. Launch windows can slip, so 'June' should be read as a target rather than a guarantee.
“Launch windows can slip, so 'June' should be read as a target rather than a guarantee.”
The stakes are competitive. With Anthropic reportedly capturing the majority of new enterprise AI spend and OpenAI pushing hard on deployment, Google needs a frontier-class Pro model to defend its position. A large context window and strong reasoning are exactly the capabilities enterprise buyers weigh when standardizing on a default model.