
Google launches Gemini 4 Hydrogen for Search, the app and developers; free users get it via «Auto»
Redacción IvokaNov 13, 9:00 AM 3 min
Google's new «everyday» model began rolling out on November 12 in Search, the Gemini app and for developers. The speed Google reports is measured only against Gemini 3.8 Flash.
Google launched Gemini 4 Hydrogen on November 12, 2026, beginning a global rollout of a model designed for fast coding and agentic work. It joins Gemini 4 Carbon and Argon, which focus on deeper reasoning and longer tasks.
Where you'll see it
- Google Search. Rolling out globally to all users.
- Gemini app. It appears in the app's model menu. It's labeled «Everyday help».
- If you use Gemini for free. You can't pick it directly: you get it through automatic model selection when Google deems it right for your prompt, and most free prompts will still be handled by a Flash-Lite model.
For people who code or build apps
It's in Google AI Studio and the Gemini API on every tier, at $1.50 per million input tokens and $7.50 per million output tokens. A «token» is roughly a piece of a word: that price is what developers pay to use the model in their apps.
What Google says about speed
According to Google, Hydrogen finishes tasks 2.5x faster than Gemini 3.8 Flash at its default («high») thinking setting. That comparison is only against Google's own previous model, not against other companies' models.
An independent test cited by iGeeksBlog, run by Theo (t3.gg) on a single code review task, painted a different picture against rivals: Hydrogen took 4 minutes 43 seconds, faster than Gemini 4 Carbon (14 minutes 8 seconds) but slower than Opus 5.5 (2 minutes 53 seconds) and GPT-6.1 Sol (2 minutes 17 seconds). It's one task, but it shows that being faster than Carbon or 3.8 Flash doesn't mean being faster than every rival.
Note that most of the numbers come from Google's own tests or from partners Google doesn't name. The model card also acknowledges that on a terminal-based coding test (Terminal-Bench 4.1, cheating-adjusted), Claude Sonnet 5.5 scores 63.9%, above Hydrogen's 57.0%.

