Google released Gemini 4 Hydrogen on 12 November, positioning it as the company's daily driver for everyday coding and agentic workflows rather than another frontier reasoning model. Where Gemini 4 Carbon and Argon are tuned for long-horizon problem solving, Hydrogen is built for high velocity. Google says it runs at 2.5x the speed of Gemini 3.8 Flash and ships with four selectable thinking levels, from low through a new xhigh tier, so developers only pay for the reasoning depth a task actually needs.
The benchmark table Google DeepMind published with the launch is unusually blunt. Hydrogen outscores Argon on the Artificial Analysis Intelligence Index at its default high thinking level, 51 versus 50, and climbs to 54 at xhigh, while edging Claude Sonnet 5.5 and GPT-6.1 Terra on agentic coding, knowledge work, multimodal understanding and reasoning. Anthropic's model still takes computer use and parts of the science suite. Pricing is flat across the entire one million token context window at $1.50 per million input tokens and $7.50 per million output tokens.
Hydrogen is rolling out globally the same day. Developers get it through the Gemini API in AI Studio and Android Studio, plus client-side via Firebase AI Logic. Enterprises get it in Gemini for business, Google Antigravity and Gemini in Sheets for eligible subscriptions. Consumers get it in the Gemini app model dropdown and in Google Search, where it has already become the default model for AI Mode worldwide. Robby Stein, VP of Product for Google Search, said the model is also handling the hardest questions in AI Overviews, with better results on multi-part and typo-heavy queries.