MOUNTAIN VIEW, Calif. — Google has launched Gemini 4 Hydrogen, a new artificial intelligence model it describes as its fastest yet, rolling it out globally across Google Search, the Gemini app, Google AI Studio, the Gemini API and its business tools.
Hydrogen, announced on Wednesday, joins Carbon and Argon in the Gemini 4 family. Google positions it as the model for fast execution across everyday and multistep tasks — coding, debugging, research, data analysis and web development — while Carbon and Argon are aimed at deeper, longer-running reasoning.
Google says Hydrogen completes tasks 2.5 times as fast as Gemini 3.8 Flash at its default high thinking setting. It scored 57 percent on Terminal-Bench 4.1, a coding benchmark, compared with 18.2 percent for Gemini 3.8 Flash, and Google says it reduces fabricated claims by more than five times. The model supports a context window of up to 1 million tokens and is priced at $1.50 per million input tokens and $7.50 per million output tokens.
In Search, Hydrogen is the default model in AI Mode and handles harder questions in AI Overviews, according to the company's announcement. Google says the model better understands questions asked in natural language — including multi-part queries, typos and highly specific requests — and delivers more visual answers, with images and comparison tables, while connecting users to a broader range of links through an upgraded query fan-out technique.
For consumers, Hydrogen is available through the model menu in the Gemini app, with free users getting it through automatic model selection, and in Google Search for all users globally. Developers can use it through the Gemini API and Google AI Studio, and paid Google AI Pro and Ultra subscribers get it in Google Antigravity, the company's agentic development platform.
The launch comes as Google faces intensifying competition in AI search and coding assistants. The company calls Hydrogen its biggest model upgrade to Search yet — and, with the model now answering the hardest questions for all users, the quality of its answers will be judged in public, one query at a time.
