Google launched Gemini 4 Hydrogen worldwide on November 12, 2026. The new AI model is rolling out to its apps, Search and tools for coders. The company says Hydrogen focuses on fast everyday work, while Carbon and Argon handle longer, harder tasks.
Google’s launch includes free users, but access depends on the product. Its help page says the app can choose Hydrogen for free users. But they cannot pick it for every chat.
What Gemini 4 Hydrogen changes in Search
Google says Search can better handle questions with several parts, specific needs or typing errors. It also promises more images, tables and links in responses across supported AI Mode languages.
The Search team’s announcement explains how this works: the system runs several searches across related parts of a question. Google calls this “query fan-out.” The aim is to find useful sites for each part of the request, rather than just one.
Free access comes with conditions
According to Google’s help page, most free-tier prompts still use Flash-Lite, a lighter model for quick answers. The app may send harder prompts to Hydrogen. Turning that option off keeps all responses on Flash-Lite.
The same guide says access depends on the user’s plan and how much they have used it. Work and school accounts follow their Workspace license, so the personal-account rules do not apply to every user.
Gemini 4 Hydrogen costs and checks
Google’s guide for coders says Hydrogen can write code, find bugs and build websites. Coders can use it in AI Studio and Android Studio through the Gemini API. This is a way for software to call the model. Google Antigravity access requires a paid AI Pro or Ultra subscription.
Google also launched a tool called Orchestrator in Antigravity. It can switch models within one work session. Google says Carbon can plan harder work, while Hydrogen can carry out tasks. That adds a different role alongside the earlier Gemini 4 Carbon release.
The test group Artificial Analysis gives Hydrogen’s High setting a score of 51 on its Intelligence Index. That was its score as of November 13, 2026. Its model page lists prices of $1.50 per million input tokens and $7.50 per million output tokens. These are API rates, not Gemini app subscription prices.
Artificial Analysis says its index draws on nine tests. They cover knowledge work, coding, science and general knowledge. It advises checking individual tests for a specific task. A single overall score therefore gives only part of the picture.
Long files still have limits
Google’s token guide explains that text is split into units that can be words or parts of words. Those units help determine API charges. The Gemini 4 Hydrogen model card adds a key detail for large files. Input and output share the one-million-token limit. For a reply sent straight back to the calling software, it lists a default limit of 64,000 output tokens. A process that runs in the background can return up to one million tokens.
The card also says Hydrogen can invent information and may slow down or time out. It lists a March 2026 knowledge cutoff, with some areas less current. At launch, Google said a fuller safety and security report was still to come.
Sources: Google token guide, Google launch, Google Search, Google access guide, Google developer guide, Google model card, Artificial Analysis results, Artificial Analysis methodology
