Alphabet’s Google has released Gemini 4 Hydrogen, a faster AI model aimed at everyday coding and business tasks. The launch brings Gemini 4 to every user tier, from free consumers to large companies.
Pricing is a central part of the release. Hydrogen costs $1.50 per million input tokens and $7.50 per million output tokens, with the same rate across its full 1 million-token context window.
Google’s larger models, Gemini 4 Carbon and Gemini 4 Argon, remain its tools for deep, long reasoning on hard problems. Hydrogen is built for quick execution, such as writing code, running tests and checking changes.
A Faster Model That Scores Higher
On the Artificial Analysis Intelligence Index, Hydrogen scored 51 at its default high thinking level. That edges out Argon at 50, and Hydrogen reached 54 at its new top “xhigh” setting.
The model offers four thinking levels: low, medium, high and xhigh. Users can pick how much reasoning a task needs, which affects both wait time and cost.
Compared with Gemini 3.8 Flash, Hydrogen finishes tasks 2.5 times faster at its default setting. Its Terminal-Bench 4.1 score rose to 57% from 18.2%, and fabricated claims fell by more than five times.
Early business testers reported similar gains. Box said Hydrogen ran about twice as fast as 3.8 Flash and used 12% fewer tokens, with the biggest improvements in data-heavy work like reading tables.
Pixelcut ran 276 tests and said the model completed every design brief. It reported 1.8 times faster speed and 30% cleaner code.
Orchestrator Splits Work Between Models
Google also introduced Orchestrator in its Google Antigravity platform. The tool switches between models like Carbon and Hydrogen within one session instead of using a single model throughout.
On Terminal-Bench 4.0, Orchestrator came close to Carbon’s task success rate. It used 24% less quota on average and finished 19% faster.
For larger projects, Carbon can act as the coordinator while Hydrogen runs worker and checker agents. These agents build and test changes at the same time in separate workspaces.
Hydrogen handles up to 1 million input tokens and up to 1 million output tokens. It can turn PDFs, spreadsheets, charts, audio and video into cited reports.
On safety, the company said Hydrogen cuts misaligned actions and task shortcutting by 12 times compared with 3.8 Flash. It scored a 96.7% honesty rate on an internal agent test.
The model is more than 2.3 times more resistant to indirect prompt injection attacks on the Gray Swan benchmark. False refusals on HealthBench Pro stayed under 0.4%.
In Google Search, Hydrogen is designed to better understand multi-part questions and typos. It also links users to a wider range of websites than before.
Hydrogen began rolling out globally on November 12. Developers can access it through the Gemini API, AI Studio and Android Studio, while businesses can use it in Gemini for business and Gemini in Google Sheets.
Consumers can select it in the Gemini app and use it in Google Search. Paid Google AI Pro and Ultra subscribers can also try it in Google Antigravity.
