Gemini Drops October 2026: Gemini 4 Argon, Skills and a Windows App

What this is about

On October 30, 2026, Google released the October edition of "Gemini Drops," a monthly roundup of new features in the Gemini app. It covers eleven items. The details come from the Japanese blog Jetstream, which draws on Google's announcement, so they reflect availability in Japan. Seven of the eleven features can be used there. Four are so far available only in English-language environments. The sources don't say what the situation is in Germany.

Gemini 4 Argon

The headline item is Gemini 4 Argon, which Google calls the first model of the new "Gemini 4" generation. It breaks complex tasks into steps and is meant to turn ideas into finished results, such as shareable web apps, 3D files, and polished slides and reports. Google AI Pro and AI Ultra subscribers worldwide get access in the Gemini app and in Search's "AI Mode."

New features at a glance

Details on individual features

Antigravity on the Mac: A button in the macOS app takes you from the chat into Google's autonomous coding environment, Antigravity. There you can build, run and debug software, prototype 3D games and automate workflows on your Mac. Once you approve the plan, the agent works in the background and checks in for review. English only, ages 18 and up.

Windows app: Available since September 10, 2026. Press Alt+Space to call up Gemini over your current work to summarize documents, brainstorm or draft text. It runs in every country and language where the web app is supported.

Skills: Instructions you use often are saved once and reused. You can create them in a chat with Gemini and combine several. They replace Gems; from November 17, 2026, Gems are set to be migrated automatically.

Personal Intelligence: With your permission, Gemini can link your Maps search history and saved places with calendar events. For example, if you ask where you could eat after your last meeting, Gemini looks for saved restaurants nearby.

Gemini Notebook and connections: With "@Gemini Notebook," you add content to notebooks or ask questions about it. The integration is on by default (off in the EEA). More than 40 services can be connected, and the feature has been called "Connectors" since mid-October. Most of them apply only to English in the US.

Files: You can create, view and edit PDFs, Google Docs and Sheets in chat. On mobile and the web, only PDF is supported; other formats work only in the web app.

Lyria 3.5: In the app since September 4, 2026. Pick a genre, set the length, with or without vocals. Google calls it its most sonically powerful music model yet. Available worldwide to all users.

Avatars: Shared avatars let you create images and videos together with a similar look and voice. You can revoke access at any time. English only for now.

Free access is shrinking

Notebookcheck reports that, according to Google's help article "Changes to Gemini model access and limits," users without a subscription can only use Flash-Lite from October 9, no longer Flash and Pro. AI Plus (US$4.99 in the US) loses Pro, while AI Pro ($19.99) and Ultra keep all models. Pro subscribers get Deep Think, though according to the limits page, so far only Ultra does. Each model gets an effort level (low, medium, high), and higher levels use up more of your limit. Some Google pages haven't been updated yet, and the Notebookcheck article points out contradictions.

Our take

The Drops show where Google is headed: agents, integration into desktop and services, and personalization. The strongest new features, such as Gemini 4 Argon, are mainly open to paying customers, while free access is shrinking. It's unclear how well Gemini 4 Argon performs in benchmarks, since the sources give no numbers, and when the English-only features will arrive in other countries. The horoscope sources don't belong to this story and were ignored.