Google Updates

Everything Google has shipped, in one place.

This Week

Googlefeature

Link·Sep 10, 2026

Gemini App for Windows

Google's Gemini desktop app, already on macOS, now runs on Windows 10 and 11 and opens with Alt + Space from any app.

Best for: Quick drafting, summarizing, and image or video creation without leaving the app you're working in
When to use it: You keep a Gemini browser tab open all day and switch to it every few minutes.
Who benefits: Windows users who work across Gmail, Drive, and everyday desktop apps

Googlefeature

Link·Sep 8, 2026

Gemini 3.5 Transcribe in the Gemini App for macOS

Speak to Gemini on a Mac and have it act on what is already open on your screen.

Best for: Working through files and research by talking instead of typing.
When to use it: What you need Gemini to act on is already open on your screen and describing it in a prompt is slower than pointing at it.
Who benefits: Mac users who do research and planning work in long sessions.

Googlefeature

Link·Sep 8, 2026

Quiz Follow-Ups in Gemini Notebook

After a quiz, Gemini Notebook can tell you what to study next or build flashcards aimed at what you got wrong.

Best for: Turning a pile of source documents into a study loop that targets your actual gaps.
When to use it: You have taken a quiz on your own material and want the next round aimed at what you missed.
Who benefits: Anyone onboarding to a dense body of material, including marketers learning a new product or category.

Last Week

Googlemodel

Link·Sep 3, 2026

WeatherNext 3

Google's weather model, now producing hourly forecasts at 5-kilometer resolution and powering weather across Search, Maps, and the Gemini app.

Best for: weather data precise enough to plan around at a neighborhood level rather than a city level
When to use it: you build something that depends on forecast accuracy, or you want better weather answers inside Google's own products
Who benefits: developers pulling weather data through Google Cloud, plus anyone using Google Search, Maps, or the Gemini app

Googlemodel

Link·Sep 3, 2026

Lyria 3.5 (Public Preview)

Google's music generation model, now able to produce full-length songs through the Gemini API.

Best for: generating original music for video, ads, or product experiences without licensing a track
When to use it: you need music shaped to a specific length and structure rather than a generic loop
Who benefits: developers building media tools, and creative teams producing audio at volume

Googlepricing

Link·Sep 2, 2026

Flexible Usage Limits in Gemini Notebook

Gemini Notebook replaces a separate daily cap for each artifact type with one allowance you spend where you want.

Best for: getting more out of a plan when your work leans heavily on one kind of output
When to use it: you keep hitting the cap on one artifact type while the others go unused
Who benefits: Gemini Notebook users who work in bursts rather than evenly across a day

Googlemodel

Link·Sep 2, 2026

Gemini 3.8 Flash Cyber

A version of Gemini 3.8 Flash tuned for security defenders, available only to vetted organizations.

Best for: finding software vulnerabilities and generating the patches that fix them
When to use it: a security team needs automated review across a large codebase rather than spot checks
Who benefits: vetted defenders in Google's Fairwind Program, including government bodies and critical-infrastructure operators

Googlemodel

Link·Sep 2, 2026

Gemini 3.8 Flash

Google's new mid-tier Gemini model, built for long multi-step tasks and available to Google AI Pro and Ultra subscribers.

Best for: Tasks with many steps that need to hold together, like working through a research question or running a multi-stage workflow
When to use it: A job is too involved for a quick answer but does not warrant top-tier model pricing
Who benefits: Google AI Pro and Ultra subscribers, and developers building agents on the Gemini API

Googlefeature

Link·Sep 1, 2026

Short Video Overviews in 70+ Languages

Gemini Notebook can now turn your sources into a roughly 60-second vertical video in more than 70 languages.

Best for: Turning a set of documents into a short vertical video for a non-English audience
When to use it: You need the same explainer in several languages and do not want to produce each one separately
Who benefits: Google AI Ultra and Pro subscribers doing internal comms or content for multi-language teams

Googlefeature

Link·Sep 1, 2026

Agentic Video Understanding

Gemini now hunts through a video for the parts it needs instead of reading every frame, cutting the token cost of video work by up to 88%.

Best for: Pulling specific moments or answers out of long video without paying to process the whole thing
When to use it: Your source is hours of footage, a webinar, or a recorded call and you only need parts of it
Who benefits: Teams working with video content, recorded calls, or long-form footage through the Gemini API