Google pushes Gemini 3.5 Pro to mid-July, and says the wait is deliberate

Google announced this week that Gemini 3.5 Pro, the flagship model it unveiled at I/O back on May 19, will not reach general availability in June as originally planned. The new target is July 17.

What changed

Gemini 3.5 Pro was announced alongside Gemini 3.5 Flash at I/O 2026. Flash shipped quickly and is already generally available — it now powers the Gemini app and much of Google's agent push. Pro, the larger and more capable sibling, was supposed to follow in June.

Instead, Google says the model remains in internal testing and a limited enterprise preview. The official explanation is that the team is refining performance on complex tasks after feedback from early enterprise testers. Some reporting goes further, suggesting Google DeepMind opted for a substantial architectural rebuild rather than iterating on the older 2.5 Pro foundations — a bigger engineering lift that pushed the date.

What Gemini 3.5 Pro is expected to bring

Based on what Google has previewed and what has been reported, the model is expected to feature a context window of around two million tokens, a deeper reasoning layer for hard multi-step problems, and stronger autonomous workflow abilities — meaning it can carry out longer chains of work without a person approving each step.

The delay also happens to land Gemini 3.5 Pro in a very crowded window. OpenAI's GPT-5.6 family and new releases from other frontier labs are all converging on mid-July, so Google will be launching straight into a head-to-head comparison.

A delay is not a stumble

It is tempting to read every delay as trouble, but the recent pattern across the industry suggests otherwise. Labs have been burned by shipping models that benchmark well and then disappoint in real work. Holding a flagship back a few weeks to fix behavior on complex tasks is arguably the more mature move — especially when your fast, cheap model is already out and earning goodwill.

Why it matters for small business

Two practical points. First, if you were waiting on Gemini 3.5 Pro before committing to a Google-based workflow, you now have a firm date to plan around — and Gemini 3.5 Flash is already available and is more than enough for most everyday business tasks like drafting, summarizing, and spreadsheet help. Second, the mid-July pileup of flagship releases means pricing pressure. When three or four top-tier models launch within weeks of each other, per-token prices and subscription perks tend to improve fast. If you are about to sign an annual AI contract, it may literally pay to wait three weeks and compare.