At Google I/O in May, Sundar Pichai stood on stage, announced Gemini 3.5 Pro — and then asked the audience to wait another month. Reportedly there were audible groans in the room. Now, in early June, that month is almost up, and the launch is getting close.
What Gemini 3.5 Pro is supposed to do
The specs are ambitious: a two-million-token context window, a dedicated Deep Think mode for hard reasoning, and what Google calls frontier multimodality — working seamlessly across text, images, and other formats. On paper, that aims straight at Claude Opus and GPT-5.5.
There’s just one catch
Gemini 3.5 Pro still isn’t available. As of late May it was running in a limited preview for select Vertex enterprise customers. General availability is targeted for June — but there’s no firm date. The smaller variant, Gemini 3.5 Flash, has been generally available since May and is already in production. The Pro model, the actual heavyweight, keeps people waiting.
My take
Announce it, then make people wait — that’s risky. In a market where Anthropic ships new Claude Code versions weekly and OpenAI keeps countering, a month of waiting feels long. Google has the infrastructure, the cloud, and the balance sheet to win on price. But reach doesn’t replace a model you can actually use.
I’m curious whether those two million tokens hold up in practice — huge context windows sound great until you notice the model loses the thread halfway through. If Google nails that, things get serious for Anthropic and OpenAI. If not, it’s another impressive number on a slide.
Sources: TechTimes, Google Blog