Google’s Gemini AI division has been on a roll lately. Merely weeks after introducing the Gemini 4 Carbon and Gemini 4 Argon frontier AI models, the company has today launched Gemini 4 Hydrogen. In an official blog post, Google describes the new Hydrogen model as its daily driver for agentic workflows, with a focus on speed and efficiency, while Carbon and Argon handle deep, long-horizon reasoning.
Starting today, Gemini 4 Hydrogen is rolling out in the Gemini app, and powers Google Search for all users globally. It has also been deployed in Google AI Studio, and in Google Antigravity for users who pay for a Google AI Pro or Ultra subscription.
Carbon and Argon do the deep thinking. Hydrogen gets the everyday work done, fast.
Google’s benchmarks suggest some impressive advancements for Gemini 4 Hydrogen at agentic coding tasks. It ranks far ahead of OpenAI’s GPT 6.1 Terra and Anthropic’s Claude Sonnet 5.5 on knowledge-based benchmarks, where Gemini 4 Hydrogen is once again comfortably ahead of its rivals.
At tasks that involve multimodal understanding, such as images and videos, Google’s latest Gemini 4 series model is once again touted to be ahead of its Anthropic and OpenAI competitors. It only trails behind them in science and math benchmarks, computer use, and Terminal-Bench 4.1, though the gap is pretty small.
Compared to Gemini 3.8 Flash, Google claims that Gemini 4 Hydrogen can perform tasks 2.5x faster at its default setting, while making more than 5x fewer fabricated claims. Interestingly, Google says that in partner tests involving complex graphs, SQL, and data-heavy enterprise work, Hydrogen was up to 2.8x faster, while using up to 23% fewer tokens.
Furthermore, Google is promising some impressive gains at agentic and multimodal knowledge work, such as working with PDFs, spreadsheets, and charts, or turning audio and recorded video into reports. Gemini 4 Hydrogen lets users pick between four thinking levels, with xhigh being the newest addition.
In the Gemini app, the new model is claimed to offer a much better experience when it comes to research, writing, and daily chores like handling emails, documents, study guides, and more. It’s also said to be more efficient at picking the right tools to get the job done, such as pulling context from connected services like Gmail and Google Drive to answer your queries and execute tasks.
Hydrogen also gives a big boost to Google Search, with a better understanding of what users are asking, more visual responses, and a broader selection of links, so that users can discover relevant content across different websites.
On the topic of safety, Google claims that Gemini 4 Hydrogen cuts down on instances of AI overreach in agentic tasks by up to 81%, and is up to 2x less likely to bypass permission boundaries when a task is blocked.
For developers, Gemini 4 Hydrogen is now available through the Gemini API, including in Android Studio, and in Google AI Studio across all tiers. Enterprises can deploy it via the API or pick it from the model drop-down in Gemini Enterprise. For everyone else, it’s available in the model drop-down picker within the Gemini app, and free users will get it automatically when the app decides it’s the right fit for a prompt. It’s also coming to Gemini in Google Sheets, starting with Google AI Pro and Ultra subscribers.