Google is rolling out three new variants of its large language model, Gemini: Gemini 3.6 Flash, Gemini 3.5 Flash-Lite and Gemini 3.5 Flash Cyber. According to the Mountain View company, the main goal is to deliver better answers in the shortest possible time. But that is not all, because there is also a strong focus on energy efficiency.

Based on the earliest data available, these new models can consume roughly one-sixth fewer tokens than their predecessor, Gemini 3.5 Flash. At the same time, the tokens themselves are becoming cheaper. Still missing is the Pro version of Gemini 3.5, designed for more complex tasks.

The three new Gemini models

Let’s start with Gemini 3.6 Flash, the newly launched model built to respond to user requests as quickly as possible. Its main capabilities include writing code, studying and analyzing data, running automations, and searching the web.

Next comes Gemini 3.5 Flash-Lite, an even lighter alternative. Here, Google’s developers designed a model able to handle simpler tasks at high speed. If, for example, you want to ask Gemini a quick question or need clarification on a given topic within seconds, 3.5 Flash-Lite is the best choice.

Finally there is Gemini 3.5 Flash Cyber, a variant designed exclusively to strengthen security. This too is a lightweight model, but one whose protective shields make it suitable for more sensitive requests, such as managing security operations or analyzing an internal system.

All Gemini Advanced subscribers will be able to use the three models with a wider context window — an advantage that allows them, for instance, to upload large files for analysis or to request a project to be carried out in multiple stages.

Where is Gemini 3.5 Pro?

Gemini 3.5 Pro remains shrouded in mystery. Google’s most advanced model, whose launch was expected in these very weeks, is slow to arrive. In fact, several sources confirm that there have been development delays that could push the rollout back even further.

The situation appears to be complicated by the fact that Google is integrating its artificial intelligence into many services and applications at once — whether Google Maps, YouTube or Gmail. What does that mean? That more Google teams have a say, and each may require additional changes to work better.

This helps slow down the entire production process. In the announcement released in the past few hours, Google stated that testing of Gemini 3.5 Pro is already under way with its partners, and a wider release is expected as soon as everything is ready. In the meantime, the company says it is gearing up to train the next model, Gemini 4.

Editor’s note

This article was originally published in Italian on money.it by Pasquale Conte on July 22, 2026 as «Google Gemini, arrivano 3 nuovi modelli. Ecco tutte le novità». It has been translated and adapted for an international audience by the Money.it International desk.