Google shipped a new AI model on November 12, and it went straight into the product most people actually use. Gemini 4 Hydrogen is now the default model in Search AI Mode, Google said in its announcement, and it handles the hardest questions in AI Overviews too. The model costs $1.50 per million input tokens, and Google claims it runs 2.5 times faster than the model it replaces.

That last number matters more than the name. Hydrogen takes over from Gemini 3.8 Flash as Google’s fast model, and it lands in Search for all users, not just developers or paying subscribers. Free Gemini app users get it through automatic model selection, while developers can call it directly through the Gemini API and Google AI Studio.

Google built Hydrogen for speed over depth. It’s the model for everyday coding and agentic work — the jobs where an AI carries out several steps on its own, like researching a topic, comparing options, or navigating a codebase. Four thinking levels — low, medium, high, and a new extra-high “xhigh” — let users trade a little reasoning for a little more speed, or the other way around.

Search is where most people will meet Hydrogen first, and Google is pitching the change as a big one. Robby Stein, a Google vice president who leads Search, called it “the biggest model upgrade to Search yet,” according to The Verge.

The practical difference shows up in messy questions. Google says Hydrogen handles queries with multiple parts, highly specific requests, and typos that used to trip Search up. It also issues several queries on related topics at once, then pulls the results together. Answers in AI Mode come back with more images, structured responses, comparison tables, and custom tools, plus links Google says are more helpful than before.

That is a real shift in how Search works, not just a faster engine under the same hood. AI Mode used to be the place Google put its newest model behind a tap. Now the newest fast model is the default, and the hardest AI Overviews questions get routed to it as well.

The numbers Google is proud of

Google’s own benchmarks put Hydrogen ahead of its bigger sibling on at least one test. At its default high thinking setting, Hydrogen scores 51 on the Artificial Analysis Intelligence Index, against 50 for Gemini 4 Argon. At xhigh, it reaches 54.

On Terminal-Bench 4.1, a test of real coding tasks in a terminal, Google says Hydrogen scores 57%, up from 18.2% for Gemini 3.8 Flash. For multi-step work like deep research, data analysis, and broad codebase navigation, Google says Hydrogen ranks fifth on Agent Arena’s leaderboard.

Take those figures the way you’d take any vendor benchmark: as a claim, not a verdict. Google picks the tests it runs, and independent checks will tell a fuller story in the weeks ahead. Still, the direction is clear. A smaller, cheaper model is now matching or beating the company’s flagship on the tests Google chose to publish.

What Gemini 4 Hydrogen costs, and the free tier

Hydrogen is the cheapest model in the Gemini 4 family to run. Google prices it at $1.50 per million input tokens and $7.50 per million output tokens, flat across its entire 1 million-token context window. Batch and Flex API calls cost half that.

The bigger news for developers is the free tier. Unlike Carbon and Argon, Hydrogen is free of charge on the Gemini API free tier, in Google AI Studio, in Search, and in the Gemini app through automatic selection. Carbon and Argon cost $2 and $10 per million tokens until November 30, then double to $4 and $20 on December 1.

Cheap tokens don’t always mean cheap tasks, though. Hydrogen thinks even at its low setting, and thinking tokens count as output, so a long job can still add up. The real test is cost per finished task, not cost per token, and developers will work that out by running real workloads against it.

Where you can try it

Hydrogen is rolling out globally across Google’s developer, enterprise, and consumer surfaces: Google AI Studio, the Gemini API, the Gemini app, Gemini for business, and Google Search. Google AI Pro and Ultra subscribers also get it in Antigravity, Google’s coding platform.

It is the third Gemini 4 model in about six weeks. Argon arrived September 30, Carbon went generally available November 10, and now Hydrogen completes the set. Each one has a job. Argon is the everyday workhorse, Carbon takes the hardest tasks, and Hydrogen is the fast, cheap one that lives where most people will actually run into it.

Why this matters

Google’s model releases used to land in a developer console first and reach consumers months later. Hydrogen flipped that order. It launched in Search on day one, for everyone.

That’s the tell. The race between Google, OpenAI, and Anthropic is no longer just about who has the smartest model on a benchmark chart. It’s about whose model answers the question you typed into a search box, and whether it does it fast enough and cheaply enough that the company can afford to give it away.

Hydrogen is Google’s bet that the answer is a smaller model, priced low, running everywhere at once. If developers find it good enough for real work — and if Search answers actually get better for ordinary users — the flagship models start looking like a luxury most people don’t need.

Frequently asked questions

What is Google Gemini 4 Hydrogen?

Gemini 4 Hydrogen is Google’s fast, low-cost AI model for everyday coding and agentic work. It launched November 12, 2026, and is now the default model in Google Search AI Mode.

How much does Gemini 4 Hydrogen cost?

It costs $1.50 per million input tokens and $7.50 per million output tokens, flat across its full 1 million-token context window. It is also free on the Gemini API free tier.

How fast is Gemini 4 Hydrogen?

Google says it completes tasks 2.5 times faster than Gemini 3.8 Flash at its default high thinking setting. It also scores 51 on the Artificial Analysis Intelligence Index, one point ahead of the larger Gemini 4 Argon.

Where can I use Gemini 4 Hydrogen?

It is rolling out in Google Search, the Gemini app, Google AI Studio, the Gemini API, Gemini for business, and Antigravity for Google AI Pro and Ultra subscribers.

Sources: Google DeepMind blog, The Verge, TechCrunch