Google has launched Gemini 4 Hydrogen, a new artificial intelligence model built for speed and everyday agentic work, and has started rolling it out across the Gemini app, Google Search, its developer tools and its business products.

Google announced the model on 12 November 2026, describing it as the fast execution member of the Gemini 4 family. While its frontier models, Gemini 4 Carbon and Gemini 4 Argon, are designed for deep, long-horizon reasoning on difficult problems, Hydrogen is aimed at the everyday work that sits between those hard tasks: writing code, running tests, checking changes, carrying out research and completing multi-step jobs.

Google calls Hydrogen its daily driver for agentic workflows — the model developers and users reach for first, with larger models called in for deeper reasoning.

What Hydrogen is built to do

Hydrogen is designed for rapid execution across varied agentic workflows. In software development, Google says it handles repository coding and debugging, interactive web development and multimodal work, including reading dense documents and pulling together information from different sources.

For developers moving from Gemini 3.8 Flash, Google reports a substantial step forward. It says Hydrogen finishes tasks 2.5 times faster at its default high thinking setting, while its score on Terminal-Bench 4.1, a benchmark for terminal-based coding tasks, rises to 57 percent from 18.2 percent for 3.8 Flash. Google also says the model cuts fabricated claims by more than five times compared with its predecessor.

On the Artificial Analysis Intelligence Index, Google reports that Hydrogen scores 51 at its default high thinking level, compared with 50 for Argon, and reaches 54 at its highest thinking setting.

Four thinking levels and flat pricing

Hydrogen offers four thinking levels — low, medium, high and a new xhigh tier — so users can trade reasoning depth against speed and cost. The high level is the default. Google says the levels scale in a smooth progression, so users only wait and pay for the reasoning depth a task requires.

Pricing is flat across the model’s entire one-million-token context window: $1.50 per million input tokens and $7.50 per million output tokens. The one-million-token input window is paired with up to one million output tokens, which Google says lets the model synthesise dense PDFs, spreadsheets, charts, audio and recorded video into cited reports and structured documents.

Google is also introducing Orchestrator in Google Antigravity, its developer platform. Rather than using one model for a whole session, Orchestrator switches dynamically between models such as Carbon and Hydrogen. Google says that on Terminal-Bench 4.0, Orchestrator reached near-Carbon task success rates while using 24 percent less quota on average and completing tasks 19 percent faster. Carbon can act as coordinator while Hydrogen powers worker and verifier agents in parallel.

Where Hydrogen is available

Google says Hydrogen is rolling out globally from 12 November across its developer, enterprise and consumer surfaces. Developers can build with it through the Gemini API in Google AI Studio and Android Studio, and paid Google AI Pro and Ultra subscribers can use it in Google Antigravity.

For businesses, it is available in Gemini for business through the API or model menu, in Antigravity, and in Gemini in Google Sheets for eligible business subscriptions. For everyone else, it is available in the Gemini app through the model menu and in Google Search for all users globally, with Gemini in Google Sheets starting with Google AI Pro and Ultra subscribers.

The Search upgrade is the part most people will notice first. Google says Hydrogen brings more helpful, visual and intuitive responses to Search, with a better understanding of multi-part questions and typos, more visual answers, and a wider range of links through an upgraded query fan-out technique that connects users to sites they might not otherwise have found.

Safety and reliability

Google says Hydrogen brings improvements from its latest Gemini 4 safety and security stack to its broadest audience yet. For agentic work, it reports that the model reduces overreach in agentic tasks by up to 81 percent and under-service by up to 79 percent, and cuts misaligned actions and task shortcutting across tool-use and coding evaluations by 12 times.

It also says Hydrogen is up to twice as unlikely to bypass permission boundaries when faced with impossible or blocked tasks, and records a 96.7 percent honesty rate on its internal agent honesty evaluation. Against indirect prompt injection attacks, Google says Hydrogen is more than 2.3 times more resilient than its predecessor on the Gray Swan IPI benchmark.

Carbon and Argon remain the frontier models for the hardest reasoning, while Hydrogen is the model Google wants doing the everyday work, in Search, in the Gemini app and in developers’ coding tools.

Sources: Google Blog and Digital Trends