Gemini 4 Hydrogen: fast AI for the everyday work your business runs on

Google launched Gemini 4 Hydrogen on 12 November 2026, and it is now the default model across Google AI Studio, the Gemini API, the Gemini app, Gemini for business and Google Search. It is the fast member of the Gemini 4 family, built for everyday coding and agentic workflows, the kind of work where an AI carries out several steps on its own.

What Hydrogen is

Gemini 4 has three tiers. Carbon and Argon are the frontier models, built for deep, long-horizon reasoning on hard problems. Hydrogen is the execution engine. Google describes it as built for rapid execution across everyday coding and agentic workflows: writing code, running tests, verifying changes, doing research and carrying multi-step tasks to completion. Google adapted it from Gemini 4 Carbon, and its knowledge cutoff is March 2026, according to the model card.

The model takes text, images, audio and video as input. Its API identifier is gemini-4-hydrogen.

What it costs

Hydrogen costs $1.50 per million input tokens and $7.50 per million output tokens. The price stays flat across the full 1 million token context window, so long prompts do not trigger a higher tier. Cached input costs $0.075 per million tokens through implicit caching, which applies automatically to prompts of at least 4,096 tokens.

For comparison, Argon launched at $2 per million input tokens and $10 per million output tokens, so Hydrogen costs 25 percent less than Argon's introductory price. Gemini 3.8 Flash, the model Hydrogen replaces, costs less today, but its price rises on 1 January 2027 to the same level Hydrogen charges now.

How fast, and how good

Google says Hydrogen finishes tasks 2.5 times faster than Gemini 3.8 Flash at its default high thinking setting. On Google's numbers, it scores 51 on the Artificial Analysis Intelligence Index at that setting, against 50 for Argon, and 54 at its highest thinking setting. It scores 57 percent on Terminal-Bench 4.1, a coding benchmark, against 18.2 percent for Gemini 3.8 Flash. Google also says it cuts fabricated claims by more than five times compared with the model it replaces.

Those are Google's numbers, measured on benchmarks Google chose. Independent tests will follow, and they are the ones to watch before you rebuild anything around the model. The direction is clear, though: the cheap model is now close to the flagship on everyday work.

Hydrogen ships with four thinking levels: low, medium, high as the default, and a new xhigh tier. You wait and pay only for the reasoning depth a task needs.

What it changes for a small business

First, agent loops get cheap. An agent that fires hundreds of small requests per task was expensive with a flagship model. At Hydrogen prices, the same loop costs a fraction, which makes it practical to let an agent check a form, look up a record and draft a reply without a person watching each step.

Second, coding help gets faster. A developer or a small team can ask the model to write code, run tests and verify changes in one workflow, and the 2.5 times speed gain shows up directly in how long a task takes.

Third, Search changes for everyone. Hydrogen is now in Google Search for all users, not only subscribers. Google calls it the biggest model upgrade to Search yet, and says it runs more searches across subtopics and surfaces a wider range of websites. That means the way people find your business is changing again. Content that answers a question directly, in plain language, is what this model looks for.

How to try it first

Start in Google AI Studio, where the model is available on all tiers, including the free tier. Pick one workflow you already run by hand, such as sorting incoming enquiries or drafting first replies. Run it at the low thinking level first, and compare the results with what your team produces. Raise the thinking level only when the task is hard enough to need it. Keep a person checking the output for the first month, and log the cases where the model was confident and wrong. Those cases tell you where your criteria need to be sharper.

Read the full announcement from Google: Introducing Gemini 4 Hydrogen for Agentic Workflows

If you want that workflow inside your own website, our AI website upgrade and workflow automation services are a good place to start.

Takeaway

Hydrogen is not the smartest model Google sells. It is the fast, cheap one, and that is the point. For the everyday work a small business actually runs, speed and price matter more than record scores. Try it on one workflow, measure it against your own team, and keep the human in the loop.