Google's Gemini 4 Hydrogen Is 2.5 Times Faster Than 3.8 Flash and Now Powers the Gemini App and Search

The new everyday model in the Gemini 4 family edges past Argon on an independent intelligence index, costs developers $1.50 per million input tokens and is rolling out to free and paid users worldwide on Thursday.

MOUNTAIN VIEW, Calif. - Google on Thursday released Gemini 4 Hydrogen, a faster and cheaper model in its Gemini 4 family that takes over from the “Flash” line as the model for everyday tasks in the Gemini app and Google Search. Google said Hydrogen finishes tasks 2.5 times faster than Gemini 3.8 Flash at its default setting, and it is rolling out globally starting today, according to the company’s announcement.

Hydrogen sits below Gemini 4 Carbon and Gemini 4 Argon, the larger models Google released over the past month for long, difficult reasoning work. Koray Kavukcuoglu, a senior vice president at Google DeepMind and Google’s chief AI architect, wrote that Hydrogen “is built for rapid execution across diverse agentic workflows.”

On the Artificial Analysis Intelligence Index, an independent ranking of AI models, Hydrogen scores 51 at its default high thinking level, compared with 50 for Argon, and reaches 54 at its highest setting, Google said. The model has four thinking levels, low, medium, high and a new xhigh tier, so users can trade speed and cost for deeper reasoning.

Image: Google

Against Gemini 3.8 Flash, Google said Hydrogen more than triples the score on the Terminal-Bench 4.1 coding test, 57 percent versus 18.2 percent, and cuts fabricated claims by more than five times. Google’s own comparison table shows Hydrogen ahead of Anthropic’s Claude Sonnet 5.5 and OpenAI’s GPT-6.1 Terra on most of the tests it chose, but behind Claude Sonnet 5.5 on Terminal-Bench 4.1, the OSWorld-2.1 computer-use test and a science and math benchmark.

For developers, Hydrogen costs $1.50 per million input tokens and $7.50 per million output tokens, with the same price across its full 1 million-token context window. It can also produce up to 1 million tokens of output. It is available in Google AI Studio, the Gemini API, Gemini for business and, for paid Google AI Pro and Ultra subscribers, in Google Antigravity, the company’s coding tool.

Video: Google Antigravity on YouTube

Google also introduced Orchestrator in Antigravity, which switches between Carbon and Hydrogen within a single session, handing routine steps to the cheaper model. On Terminal-Bench 4.0, Google said, Orchestrator came close to Carbon’s success rate while using 24 percent less quota and finishing 19 percent faster.

For consumers, Hydrogen appears in the Gemini app’s model menu, and free users get it through automatic model selection when it suits the prompt. Google said the model sticks more closely to a user’s requested tone and format and more often calls the right tools, including Gmail, Google Drive and connected apps such as Airtable and Dropbox.

In Google Search, Hydrogen is now available to all users worldwide. Robby Stein, vice president of product for Search, called it “the biggest model upgrade to Search yet” in a separate post, saying AI responses can now handle multi-part questions and typos, show more images and comparison tables, and link to a broader range of sites.

Image: Google

On safety, Google said Hydrogen cuts misaligned actions and task shortcutting in its tool-use and coding evaluations by 12 times compared with 3.8 Flash, and is more than 2.3 times more resilient against indirect prompt injection attacks on the Gray Swan benchmark.

Sundar Pichai, Google’s chief executive, and Google DeepMind announced the model on X on Thursday.

Screenshot: @sundarpichai on X

Screenshot: @GoogleDeepMind on X