Google has introduced Gemini 4 Hydrogen, a new artificial intelligence model built for speed. Announced on November 12, 2026, Hydrogen is designed for everyday coding and agentic workflows — work where an AI writes code, runs tests, checks the results and moves on to the next step, over and over. It is the fast execution model in Google’s new Gemini 4 family, alongside Carbon and Argon, which are built for deeper, longer-horizon reasoning. It is already rolling out globally across Google AI Studio, the Gemini API, the Gemini app, Google Search and Google Antigravity.
Summary
- The news: Google introduced Gemini 4 Hydrogen on November 12, 2026 — a fast execution model for coding and agentic workflows.
- What it is for: Writing code, running tests, verifying changes, research and multi-step tasks — the execution side of agentic work.
- Thinking levels: Four — low, medium, high (the default) and a new xhigh tier for harder tasks.
- The benchmark: Google reports Hydrogen beats Argon on the Artificial Analysis Intelligence Index at default high (51 vs 50), reaching 54 at xhigh.
- Pricing: $1.50 per million input tokens and $7.50 per million output tokens, flat across the full 1 million-token context window.
- Availability: Rolling out globally from November 12 across AI Studio, the Gemini API, Gemini for business, the Gemini app, Google Search, and paid Pro and Ultra subscribers in Google Antigravity.
A Fast Model for Execution, Not Deep Reasoning
Hydrogen is not Google’s smartest model — it is its fastest. In the Gemini 4 family, Carbon and Argon are the frontier models, built for deep, long-horizon reasoning. Hydrogen is the model that gets things done once the plan exists: writing code, running tests, verifying changes, doing research and carrying out multi-step tasks.
That split matters because an agent may call a model dozens or hundreds of times in one session. A frontier model thinking deeply about every small step would be slow and expensive; a fast model handles the routine steps while the heavier reasoning is reserved for the moments that need it.
Google Antigravity, the company’s agentic development platform, makes that split explicit with a feature called Orchestrator, which switches between Carbon and Hydrogen within a single session, handing each step to the model best suited to it.
Four Thinking Levels, Including a New Xhigh Tier
Hydrogen ships with four thinking levels: low, medium, high (the default) and a new xhigh tier. The level controls how much reasoning the model does before it answers, so users can trade speed for care: a quick code completion might run on low, while a tricky debugging session could use xhigh.
Google’s own numbers suggest the extra thinking pays off: on the Artificial Analysis Intelligence Index, Hydrogen scores 51 at default high, just ahead of Argon at 50, and 54 at xhigh. That is a narrow lead, and it comes with an important qualification: these are Google’s own numbers, not independently verified results, so independent testing is needed before the gap means very much.
Pricing and the 1 Million-Token Context Window
Hydrogen is priced at $1.50 per million input tokens and $7.50 per million output tokens, flat across its entire 1 million-token context window — no surcharge for using the full context. That matters for agentic coding, where a session accumulates large amounts of code, test output and conversation history.
A million tokens holds a large codebase or an extended agent session in one window, and flat pricing removes the worry that cost climbs as a session grows. For developers running agents that iterate on code, predictable pricing is arguably as important as the headline rate.
An agent that writes code, runs a test, reads the failure, fixes the code and runs the test again can burn through many calls in minutes. At Hydrogen’s pricing, that loop stays cheap — which is the point. Google is selling Hydrogen as the model you can afford to run constantly, not the one you save for special occasions.
Where Hydrogen Is Rolling Out
Hydrogen began rolling out globally on November 12, 2026, in Google AI Studio and the Gemini API for developers, in Gemini for business, in the Gemini app and in Google Search. Paid Google AI Pro and Ultra subscribers also get it in Google Antigravity, where Orchestrator can switch between Carbon and Hydrogen inside a session.
One wrinkle for free users: on the free Gemini app tier, Hydrogen is available through automatic model selection only — the app may choose it for them, but they cannot select it directly. The clearest way to try Hydrogen on purpose is through AI Studio, the API or a paid plan.
Google is not launching Hydrogen as a developer-only preview: it is going into the consumer app, Search and the business tier at once, which suggests Google sees fast execution models as a general-purpose layer, not a niche tool for coders.
The Honest Caveats
Hydrogen’s launch comes with caveats worth stating plainly. First, the benchmark numbers are Google’s own and not independently audited. A one-point lead over Argon on a single index is a small margin, and it may not hold up on other tests or in real-world use.
Second, Google’s Cloud documentation lists Hydrogen as a short-term general-availability release, with a minimum retirement notice of 45 days — a shorter runway than developers usually expect, so teams should plan for a possible migration to a successor.
Third, the model card notes Hydrogen can hallucinate and can occasionally be slow or time out. Its knowledge cutoff is March 2026, and it does not support RAG Engine, model tuning or the Live API.
None of that makes Hydrogen a bad model — it makes it a new one, launched fast, with limits that are documented rather than hidden.
For now, Hydrogen looks like a pragmatic addition to Google’s AI lineup: not the smartest model in the family, but the one most likely to be running in the background all day — and, if the benchmarks hold up, the default engine for the agents doing the everyday work of software development.
Related Guides
- Claude Fable 5.2 and Mythos 5.2 Release
- Google DeepMind AlphaGenome 2 DNA Model
- Apple Siri Agentic AI Agent Ternus 2027
Frequently Asked Questions
What is Gemini 4 Hydrogen?
Gemini 4 Hydrogen is a fast execution AI model introduced by Google on November 12, 2026. It is built for everyday coding and agentic workflows — writing code, running tests, verifying changes, research and multi-step tasks — and it is part of the Gemini 4 family alongside Carbon and Argon.
How is Hydrogen different from Carbon and Argon?
Carbon and Argon are the Gemini 4 family’s frontier models, built for deep, long-horizon reasoning. Hydrogen is the fast execution model: it handles the routine, repetitive steps of agentic work quickly and cheaply, while the frontier models take on the harder reasoning. Google Antigravity’s new Orchestrator feature switches between Carbon and Hydrogen within a single session.
How much does Gemini 4 Hydrogen cost?
Hydrogen costs $1.50 per million input tokens and $7.50 per million output tokens. The pricing is flat across its entire 1 million-token context window, with no surcharge for using the full context.
What are Hydrogen’s four thinking levels?
Hydrogen offers four thinking levels: low, medium, high (the default) and a new xhigh tier. Google reports Hydrogen scores 51 on the Artificial Analysis Intelligence Index at default high, ahead of Argon at 50, and reaches 54 at xhigh. These are Google’s own figures and have not been independently verified.
Where can I use Gemini 4 Hydrogen?
Hydrogen is rolling out globally from November 12, 2026 across Google AI Studio, the Gemini API, Gemini for business, the Gemini app and Google Search, plus Google Antigravity for paid Google AI Pro and Ultra subscribers. Free Gemini app users get Hydrogen through automatic model selection only, not by direct selection.
Sources: Google blog, “Introducing Gemini 4 Hydrogen for Agentic Workflows” (Nov 12, 2026); The Next Web (Nov 12, 2026); TestingCatalog (Nov 12–13, 2026).
