Google launched Gemini 4 Hydrogen on Nov. 12, describing it as a faster model for coding and tasks handled by software agents and rolling it out across Google Search, the Gemini app and developer services.
Google describes its Gemini 4 Carbon and Argon models as built for complex reasoning that takes a long time, while Hydrogen is designed to handle everyday tasks quickly. It accepts text, images, audio and video. It can take in and produce up to one million tokens of information in total.
Google lists API prices of $1.50 per million input tokens and $7.50 per million output tokens. Developers can use Hydrogen by itself to write code, run tests and check changes. They can also combine it with larger models in workflows that use several models.
Google says Hydrogen completes tasks 2.5 times faster than Gemini 3.8 Flash at its default high setting. It scored 57% on Terminal-Bench 4.1, compared with Flash’s 18.2%. These results were reported by Google and do not guarantee the same performance for every user.
In Google Antigravity, a new Orchestrator can switch between Hydrogen and Carbon during a task. Google says this setup came close to Carbon’s success rate on Terminal-Bench 4.0. Google says it used 24% less quota and finished 19% faster.
The rollout also includes Google AI Studio, the Gemini API and Gemini for business. Paid Google AI Pro and Ultra subscribers can access it in Antigravity.