Google Puts a Faster Everyday Gemini Model Into Search and Its App Worldwide

Google released a new model, Gemini 4 Hydrogen, on Nov. 12, 2026. The company said the model began rolling out globally the same day, reaching the Gemini app, Google Search, the Gemini API and every tier of Google AI Studio.

In its announcement, Google describes Hydrogen as built for speed on everyday coding and on agentic work, meaning tasks a model carries out step by step, and presents it as the upgrade for teams moving off Gemini 3.8 Flash, the company's previous fast model. The larger Gemini 4 models, Carbon and Argon, which the page describes as already available, stay the ones Google points to for deep, long-running problems.

Hydrogen is priced flat across its entire 1 million-token context window, at $1.50 per million input tokens and $7.50 per million output tokens. Google said the model has four thinking levels, from low to a new xhigh setting, so a task's cost and wait time scale only with the reasoning depth it needs.

Every performance comparison in the announcement is Google's own measurement, or Google's own choice of which outside index to cite. On the company's own figures, Hydrogen finishes tasks 2.5 times faster than 3.8 Flash at its default setting and more than triples that model's score on the Terminal-Bench 4.1 coding test, reaching 57%. Google also reports that Hydrogen scores narrowly above Argon on the Artificial Analysis Intelligence Index, a third-party ranking the announcement itself selects. No independent evaluation of Hydrogen has been published.

Google said Hydrogen is now behind Search responses for users worldwide, and that it has widened the range of links Search returns. In the Gemini app, free users will get the model only through automatic model selection, when the app judges it the right one for the prompt.