The engine behind your AI Mode listing changed on 2 September, and there is nothing for you to switch on.
Google released Gemini 3.8 Flash and a security-focused sibling on 2 September 2026 and made the general model available the same day. 3.8 Flash is the everyday model Google points at reasoning and coding, and Google says it beats the previous 3.7 Flash on its own published benchmarks. That benchmark comparison is Google's own claim, so treat it as the maker's number rather than an independent result. Trade coverage from The Register framed the release as landing only about three weeks after the last one, a fast run of updates; the announcement itself does not put a figure on the cadence.
The second model, Gemini 3.8 Flash Cyber, is built for defenders and is gated behind Google's Fairwind Program for trusted, government and critical-infrastructure teams. Most professional-services firms will never touch it, so the news that matters here is the general model.
Where the new model actually shows up
For anyone paying for Google AI Pro or Ultra, 3.8 Flash now runs across the Gemini app, AI Mode in Google Search, and Gemini in Google Sheets, all confirmed by Google. Developers reach it through the Gemini API, Google AI Studio, Android Studio and Google's enterprise platform, as 9to5Google reported. On the API, Google set an introductory price of $0.75 per million input tokens and $3.75 per million output, running until 31 December 2026, after which the standard rate doubles to $1.50 and $7.50. Teams that value efficiency over raw capability can stay put: Google confirmed 3.7 Flash "remains fully supported for efficiency-first workloads."
Your AI Mode visibility rides on a model you do not control
3.8 Flash now powers AI Mode in Google Search and Google's other AI-answer surfaces, the places where a buyer's question gets a written answer that may name, quote or skip your firm. When the model underneath changes, the way you get described can move with it, even though your website and your rankings have not touched.
There is no setting to adjust and no ranking lever to pull. The measured response is the same one that has worked all year: keep your public-facing facts accurate, current and easy for an AI to lift cleanly, so the model has something correct to say when it reaches for you. That is the everyday work of earning AI recommendations, and it is where steady AI visibility is won or lost. Watch your AI Mode impressions in Search Console over the coming weeks; if they drift, that is your signal, not a reason to panic.
Frequently asked questions
Do I need to do anything?
No. 3.8 Flash rolls out inside Google's products automatically. There is no setting to change and no ranking action to take. Keep your public information accurate and current so any Gemini model can quote it correctly.
Will this change my Google rankings?
No. This is a model update to how Gemini reasons and answers, not a change to Search's ranking systems. Your organic positions are not directly affected, though the wording AI Mode uses to describe you may shift.
What is Gemini 3.8 Flash Cyber?
A separate, security-focused version released the same day for defenders. It is gated behind Google's Fairwind Program for trusted, government and critical-infrastructure users, so it is not relevant to most professional-services firms.
Does it cost more to use?
Through the API, Google set an introductory price of $0.75 per million input tokens and $3.75 per million output, running until 31 December 2026, after which the standard rate doubles. For Pro and Ultra subscribers using the Gemini app, it is included in the existing plan.
Should I be worried about the change?
No. The calm response is to keep your public-facing facts accurate and watch your AI Mode impressions in Search Console over the coming weeks. Nothing here needs an urgent fix.
