Google has launched Gemini 4 Hydrogen, the fast and cheaper model in its new Gemini 4 family. It is built for everyday coding and multi-step agentic work — the kind of tasks where you want a model that is quick, capable and cheap enough to run all day.
Hydrogen arrived on Thursday 12 November 2026, just two days after Google made its flagship Gemini 4 Carbon model generally available. The two models are not the same thing: Carbon is the frontier model, while Hydrogen is the fast one. Here is what Hydrogen costs, what it can do, and who should try it.
What is Gemini 4 Hydrogen?
Hydrogen is Google’s speed-first Gemini 4 model. In Google’s announcement, Google describes it as bringing cutting-edge intelligence to everyone, with speed and efficiency for everyday coding and agentic workflows — jobs where an AI agent works through several steps on its own, such as writing code, checking it, and fixing its own mistakes.
It is available across all tiers of Google AI Studio and the Gemini API, as well as in Gemini for business, the Gemini app, Google Search, and for paid Google AI Pro and Ultra subscribers in Google Antigravity. Google also says it is rolling out to the Gemini app and Google Search worldwide for all users.
The model ID for developers is gemini-4-hydrogen. It accepts text, images, PDFs, tables, audio and video as input.
Price: flat, and cheaper than the flagship
Hydrogen costs $1.50 per million input tokens and $7.50 per million output tokens, and the price is flat across its entire 1 million-token context window. There is also a free tier with limits in AI Studio.
That flat pricing matters. Some models charge more once you pass a certain context size, so long documents and big codebases get expensive fast. With Hydrogen, the price per token is the same whether you send a short prompt or fill the whole context window.
For comparison, our guide to Gemini 4 Carbon covers the frontier model, which launched at $2 per million input tokens and $10 per million output tokens as an introductory price until 30 November, rising to $4 and $20 after that. Hydrogen undercuts it on price, as a fast model should.
How fast and how smart is it?
On the Artificial Analysis Intelligence Index, a composite score for knowledge, reasoning, coding and professional tasks, Hydrogen scores 51 at its default high thinking level and 54 at its highest xhigh setting. That puts it ahead of Argon, which scores 50, while often finishing tasks faster than other models in its class.
The speed gap is clearest against the model many developers used before it. Google says Hydrogen completes test tasks about 2.5 times faster than Gemini 3.8 Flash, and produces fabricated claims more than five times less often. On Terminal-Bench 4.1, a test of how well a model completes tasks in a terminal, Hydrogen scored 57% against 18.2% for Gemini 3.8 Flash, according to Neurohive’s benchmark breakdown.
Cost per task tells the same story. On the Artificial Analysis chart, the weighted average cost of one Intelligence Index task was $3.16 for Hydrogen, against $7.77 for Gemini 3.8 Flash.
Four thinking levels, including a new xhigh tier
Hydrogen has four thinking levels: low, medium, high, and a new xhigh tier. High is the default. The idea is simple — you only wait and pay for the reasoning depth a task actually needs.
A quick question gets the low setting and a fast answer. A hard coding problem gets xhigh, which lifts the Artificial Analysis score from 51 to 54 but can take longer and use more tokens. The steps between levels are meant to feel smooth and predictable, so you can trade speed for quality without guessing.
Gemini 4 Hydrogen vs Gemini 3.8 Flash
If you were using Gemini 3.8 Flash, Hydrogen is the model Google wants you to test next. It is faster, scores far higher on coding benchmarks, and hallucinates less — and on the Artificial Analysis cost chart, a typical task costs less than half as much.
One thing to be clear about: Google has set no date to shut down Gemini 3.8 Flash, and it remains available in the API for now. Hydrogen is the next step for Flash users, not a forced switch. If your app already works well on 3.8 Flash, there is no deadline forcing you to move.
Gemini 4 Hydrogen vs Gemini 4 Carbon
The two Gemini 4 models launched this week serve different jobs. Carbon, which went generally available on 10 November, is the frontier model — the most capable one, aimed at the hardest work. Hydrogen is the fast, cheaper one, aimed at everyday coding and agentic tasks where speed and cost matter more than maximum capability.
Both share a 1 million-token context window. The choice comes down to the job: reach for Carbon when a task needs the strongest reasoning, and reach for Hydrogen when you want a quick, capable model you can run again and again without the bill climbing.
Who should try it?
Hydrogen fits developers building coding agents and multi-step workflows, teams running AI on lots of small tasks where cost adds up, and anyone who found the flagship models too slow or too expensive for everyday use.
It also fits people who simply use the Gemini app. Since Hydrogen is rolling out in the app and in Google Search, many users will get a faster model without changing anything.
Frequently asked questions
What is Gemini 4 Hydrogen?
It is Google’s fast, cheaper Gemini 4 model, launched on 12 November 2026 for everyday coding and agentic workflows. It has a 1 million-token context window, four thinking levels, and flat pricing of $1.50 per million input tokens and $7.50 per million output tokens.
How much does Gemini 4 Hydrogen cost?
$1.50 per million input tokens and $7.50 per million output tokens, flat across the whole 1 million-token context window. A free tier with limits is available in Google AI Studio.
Is Gemini 4 Hydrogen replacing Gemini 3.8 Flash?
Google is pointing 3.8 Flash users toward Hydrogen as the next model to try, and Hydrogen is faster and cheaper per task. But Google has set no shutdown date for 3.8 Flash, and it remains available in the API for now.
What is the difference between Gemini 4 Hydrogen and Gemini 4 Carbon?
Carbon is the frontier model — the most capable Gemini 4, for the hardest work. Hydrogen is the fast, cheaper model for everyday coding and agentic tasks. Both have a 1 million-token context window.
