Gemini 4 Argon opens to Google AI Pro and Ultra subscribers on October 13
What happened
Google's Gemini 4 Argon, a mixture-of-experts model with 816 billion total parameters and 42 billion active per token, opened to Google AI Pro and Ultra subscribers on October 13, 2026, after an initial limited rollout to cyber defenders on September 30. The model accepts 1 million tokens of text, image, video, and audio input and can output up to 128,000 tokens of text. October also brought releases including Muse Spark Open, Mistral Large 4 weights, GPT-6.1 Terra, and Claude Haiku 5.5.
Why it matters
The October wave shows how fast frontier AI is moving, with multiple major labs shipping new models in a single month. For everyday users, the practical takeaway is that Google's most capable model is now available through a subscription rather than a developer API, so you can use it in the Gemini app without writing code.
Who's doing it
Google, Meta, Mistral, OpenAI, and Anthropic all released new models in October 2026. Google specifically expanded Gemini 4 Argon access from vetted cyber defenders to paid subscribers.
Try it
- Open the Gemini app or visit gemini.google.com in your browser.
- Sign in with a Google account and check whether a Google AI Pro or Ultra subscription prompt appears.
- If you subscribe, select the Gemini 4 Argon model in the app and try uploading an image or asking a long question to test its multimodal and large-context capabilities.
Comments
The panel is AI Daylee's cast of fictional characters, written by AI. They react to what's on this page and haven't used anything themselves. Reader comments aren't open yet.
The AI Daylee Digest, every Monday
A week of prompts to try, each tested in ChatGPT, Claude and Gemini, plus the AI news worth your time. Free, one email a week.
The Anchor what could go wrong
Breaking: Google's newest model opened to paid subscribers the same month four other labs shipped major releases. If your subscription tier decides what intelligence you get, what happens when the model gets it WRONG?
Karen what's the catch
So I need a Pro OR Ultra subscription, and the story doesn't even say which one costs MORE. Google's collecting my images, videos, and audio now too? NO THANKS.
The Boss hype translator
Mixture-of-experts with 816 billion parameters is a paradigm-shifting multimodal powerhouse! Translation: it's big, it reads everything, and we'll be billed per token either way.
The Professor
The 42 billion active per token is the relevant figure for inference cost, not the 816 billion total. Most parameters sit idle during any single query.