Gemini 4 Hydrogen: Google's fast model is now free in Search and the Gemini app
On November 12, 2026, Google launched Gemini 4 Hydrogen, the third model in the Gemini 4 family after Argon and Carbon. It is built for speed and agents, it is cheap on the API and, unlike its bigger siblings, you can use it right now for free in Google Search and the Gemini app. Here is what it does, what it costs and what changes for you.
Leggi questo articolo in italiano
What is Gemini 4 Hydrogen?
Gemini 4 Hydrogen is Google's new fast AI model, announced on November 12, 2026, by Koray Kavukcuoglu, SVP of Google DeepMind and Chief AI Architect at Google. In the official announcement Google describes it as a fast, cost-efficient model built for reliable agentic workflows and everyday tasks.
Hydrogen is the third model in the Gemini 4 family. The first two, Gemini 4 Argon and Gemini 4 Carbon, are frontier models, the most powerful ones, built for deep reasoning on hard problems; we covered them in Gemini 4 Argon: availability, pricing and how to use it and Gemini 4 Carbon: release, benchmarks and pricing. Hydrogen does a different job: it runs many short tasks quickly. It can work on its own or alongside the frontier models, for example with Carbon coordinating while Hydrogen handles and checks individual subtasks.
The model takes up to 1 million tokens of input and can generate up to 1 million tokens of output. A token is a chunk of text, on average a little less than one word.
Where you can use Gemini 4 Hydrogen today
The biggest difference from Argon and Carbon is that Hydrogen is available to everyone right away. According to Google:
- Google Search: live for all users worldwide, no subscription needed. Robby Stein, VP of Product for Search, calls it in the dedicated post the biggest model upgrade Search has had so far.
- Gemini app: available in the model picker; free users get it automatically when a request calls for it.
- Developers: available in the Gemini API, Google AI Studio and Android Studio.
- Google Antigravity and Gemini in Google Sheets: for paid Google AI Pro and Ultra subscribers only, and for eligible business plans.
In Search, Google says Hydrogen better understands complex questions, including multi-part ones or ones with typos, runs more parallel searches on subtopics and surfaces a broader set of links, along with more images, comparison tables and structured answers. Answers improve across all languages supported by AI Mode.
How fast and how capable: Google's numbers
The figures below are reported by Google. The company notes that some results, such as Terminal-Bench 4.1, were computed in-house.
| Test | Reported result |
|---|---|
| Speed vs Gemini 3.8 Flash | Completes tasks 2.5x faster (default setting) |
| Artificial Analysis Intelligence Index | 51 (high) vs 50 for Argon; 54 at xhigh |
| Terminal-Bench 4.1 | 57%, vs 18.2% for Gemini 3.8 Flash |
| Agents' Last Exam v1.1 | 71.6% |
| Fabricated claims | more than 5x fewer than Gemini 3.8 Flash |
Hydrogen has four thinking levels: low, medium, high (the default) and the new xhigh. The higher the level, the more the model reasons before answering, and the more time and tokens it uses.
What independent tests say
On November 13, 2026, Trending Topics reported results from Artificial Analysis, a company that runs the same tests across models. The picture is more nuanced:
- at xhigh Hydrogen scores 54, one point below Claude Sonnet 5.5 and GPT-6.1 Sol; larger models such as Gemini 4 Carbon stay on top;
- on terminal tasks (Terminal-Bench 4.0) it reaches 54%, versus 64% for Claude Sonnet 5.5 and 56% for GPT-6.1 Sol;
- it is very verbose: xhigh produces almost twice the output tokens of high for just three extra points;
- at high, medium and low, though, it is among the models with the best balance of intelligence and speed.
Gemini 4 Hydrogen API pricing
For developers, Google set one flat price across the full 1 million token context window:
| Item | Price per million tokens |
|---|---|
| Input | $1.50 |
| Output | $7.50 |
As a rough example, a request that sends 100,000 tokens and gets 10,000 back costs $0.15 + $0.075, a little over 22 cents. The price does not go up for very long documents, which makes budgeting easier if you work with archives, transcripts or product catalogs.
What it means for you
If you create content or run a business, Hydrogen changes two things starting today.
- You get a stronger model for free. Quick answers in the free Gemini app are now more capable. Try it on your repetitive work: summaries, email drafts, post ideas, customer replies. To turn those into ready-made routines, the AI Business Toolkit collects prompts and workflows for marketing, sales and support that work with any model.
- Google Search answers differently. With Hydrogen, Search explores more subtopics and shows more links: your content gets found when it answers specific questions with clear sentences and verifiable facts. The same principle applies to ChatGPT, as we saw with Businesses in ChatGPT.
- Agents get cheaper. If you use the API to automate support or route requests, a fast, low-cost model like Hydrogen makes high volumes sustainable. To set up AI customer service, start with Customer Care AI; if you mostly work with Claude, Claude Business OS shows you how to organize projects and instructions.
The practical advice stays the same: compare models on your real tasks, not on leaderboards. To dig into Google's tools, read our Gemini Omni guide, and if you want a complete path to producing content with AI, start with The AI Creator System.
Frequently asked questions
Is Gemini 4 Hydrogen free?
Yes, for normal use. Since November 12, 2026, it has been live in Google Search for all users and in the free Gemini app, which uses it automatically when needed. Antigravity and Gemini in Google Sheets require a Google AI Pro or Ultra subscription, and the API is paid.
What is the difference between Gemini 4 Hydrogen, Argon and Carbon?
Argon and Carbon are Google's frontier models, built for deep reasoning on hard problems. Hydrogen is the fast, low-cost model for everyday tasks and agents. Google reports that at its default level it beats Argon on the Artificial Analysis Intelligence Index, 51 to 50.
How much does Gemini 4 Hydrogen cost on the API?
$1.50 per million input tokens and $7.50 per million output tokens, with the same price across the full 1 million token context window.
What thinking levels does Gemini 4 Hydrogen have?
Four: low, medium, high (the default) and the new xhigh. According to Google, at xhigh it scores 54 on the Artificial Analysis Intelligence Index, versus 51 at high, but it uses many more tokens.
Read next


