
Google Gemini 4 Hydrogen: Faster AI Model Replaces 3.8 Flash in Search and Gemini App
Google’s Gemini 4 Hydrogen is the name behind a fresh surge of United States tech searches on Wednesday, 18 November 2026, days after its launch. Google introduced Gemini 4 Hydrogen on 12 November 2026 as a faster model built for agentic workflows, and it is now replacing Gemini 3.8 Flash across the company’s products. The rollout covers Google Search, the Gemini app, Google AI Studio, the Gemini API and the Antigravity coding tool.
Hydrogen is adapted from Gemini 4 Carbon, Google’s strongest model for hard, long-horizon reasoning, but it is tuned for speed rather than depth. Google describes Carbon and Argon as the models for deep reasoning, while Gemini 4 Hydrogen is built for rapid execution across varied agentic workflows, where software plans and carries out multistep tasks with little supervision. The model is also generally available in GitHub Copilot.
Gemini 4 Hydrogen — Quick Reference Summary
| Detail | What is known |
|---|---|
| Announced | 12 November 2026 |
| Replaces | Gemini 3.8 Flash across Google’s products |
| Adapted from | Gemini 4 Carbon |
| Built for | Fast execution in agentic workflows |
| Thinking levels | Low, medium, high and the new xhigh |
| Context window | 1 million tokens |
| Pricing | $1.50 per million input tokens; $7.50 per million output tokens |
| Intelligence Index score | 51 at high (default); 54 at xhigh; Argon scores 50 |
| Speed claim | 2.5 times faster than Gemini 3.8 Flash in high mode (Google-reported) |
| Accuracy claim | Fabricated claims down more than fivefold vs 3.8 Flash (Google-reported) |
| Coding claim | Terminal-Bench score more than triples 3.8 Flash (Google-reported) |
| Where it is live | Google Search AI Mode, Gemini app (Pro/Ultra), AI Studio, Gemini API, Antigravity, GitHub Copilot |
What Gemini 4 Hydrogen changes for everyday users
The most visible change is in Google Search, where Gemini 4 Hydrogen is now the default model in AI Mode. Google calls the switch the biggest model improvement it has made to Search. Robby Stein, Google’s vice president of product for Search, said the model now powers AI Mode answers for users.
In the Gemini app, the picture depends on the subscription. AI Pro and Ultra subscribers get Gemini 4 Hydrogen, while AI Plus and free users remain on the older 3.5 Flash-Lite model, according to Ars Technica. Access is also more restricted in the Gemini app and AI Studio than it is in Search, so the experience is not identical everywhere yet.
How fast is Gemini 4 Hydrogen, and how is it scored?
Google says Gemini 4 Hydrogen runs 2.5 times faster than Gemini 3.8 Flash in high mode, and that it reduces fabricated claims by more than fivefold. Its Terminal-Bench score for code generation more than triples the score of 3.8 Flash. On the Artificial Analysis Intelligence Index, Hydrogen scores 51 at its default high thinking level, one point above Gemini 4 Argon at 50 and well above Gemini 3.8 Flash at 31. At the top of its four thinking levels, xhigh, Hydrogen reaches 54.
Those figures come from Google and its benchmark partners, and they have not been independently peer-reviewed. The speed and accuracy numbers describe Google’s own testing, so real-world results on a user’s own work may differ. Even so, the gap over 3.8 Flash is the reason Google is moving its products onto Hydrogen now rather than waiting.
Gemini 4 Hydrogen pricing and who pays what
Developers pay $1.50 per million input tokens and $7.50 per million output tokens through the Gemini API, flat across the model’s full 1 million-token context window. The model is also on the Gemini API free tier under Google’s usual limits. For businesses, the appeal is a model that handles routine agent work at the lowest tier of Google’s current pricing, while Carbon and Argon stay reserved for harder reasoning jobs.
The pricing puts Hydrogen in the workhorse role. Carbon coordinates parallel agents and handles the hardest reasoning, while Hydrogen runs the working agents and checks that their changes execute correctly. That split is the core of Google’s pitch: use the expensive models only where their depth is needed, and let Hydrogen carry the everyday load. The launch lands in a busy week for US tech, where another US tech story drawing heavy searches was the recent Apple outage.
Where Gemini 4 Hydrogen is available now
Google says Gemini 4 Hydrogen is rolling out globally across its developer, enterprise and consumer surfaces. That covers Google AI Studio, the Gemini API, Gemini for business, the Gemini app, Google Search, and paid Google AI Pro and Ultra subscribers in Google Antigravity. GitHub has also made it generally available in Copilot for Pro, Pro+, Max, Business and Enterprise users, with a gradual rollout.
The rollout is immediate rather than staged, unlike the launches of Carbon and Argon. Ars Technica reports that Hydrogen replaces 3.8 Flash across Google’s products, and the model is now the default in Google Search, the Gemini app, Google Sheets and developer tools.
One more change arrived alongside the launch: Gemini Gems are retiring. Gems are now read-only, and they will disappear entirely from personal accounts on 17 December 2026. Google says Gems should automatically become Skills, its replacement, though users can recreate them manually if needed. The shift sits alongside the week’s other big tech talking points, including the Game Awards 2026 nominees announcement.
Frequently asked questions about Gemini 4 Hydrogen
What is Gemini 4 Hydrogen?
A fast AI model from Google, released on 12 November 2026 and adapted from Gemini 4 Carbon. It is built for rapid execution in agentic workflows and replaces Gemini 3.8 Flash across Google’s products.
How fast is Gemini 4 Hydrogen?
Google says it runs 2.5 times faster than Gemini 3.8 Flash in high mode. On the Artificial Analysis Intelligence Index it scores 51 at its default high thinking level and 54 at its top xhigh level, against 50 for Gemini 4 Argon.
What does Gemini 4 Hydrogen cost?
Developers pay $1.50 per million input tokens and $7.50 per million output tokens through the Gemini API, flat across its 1 million-token context window. It is also available on the API free tier under Google’s usual limits.
Who gets Gemini 4 Hydrogen first?
It is now the default model in Google Search AI Mode. In the Gemini app it is live for AI Pro and Ultra subscribers, while AI Plus and free users remain on the older 3.5 Flash-Lite model. Developers can call it through the Gemini API and Google AI Studio.
Does Gemini 4 Hydrogen replace 3.8 Flash everywhere?
Google is replacing 3.8 Flash with Hydrogen across its products, including Search, the Gemini app, Sheets and developer tools. Carbon and Argon remain available for deep, long-horizon reasoning work.
What are the four thinking levels in Gemini 4 Hydrogen?
Low, medium, high and the new xhigh. Higher levels spend more compute on a task; the default high level scores 51 on the Artificial Analysis Intelligence Index, and xhigh reaches 54.




