
Google released Gemini 4 Hydrogen on 12 November 2026, a faster everyday AI model that it recommends as the upgrade from Gemini 3.8 Flash. It is rolling out globally in the Gemini app, Google Search, Google Sheets and the Gemini API. The catch for businesses: on the paid API it costs $1.50 per million input tokens and $7.50 per million output tokens, double the current Flash rate, so test it before you switch.
What happened
Google announced the model in a post on its blog, "Introducing Gemini 4 Hydrogen", written by Koray Kavukcuoglu, Senior Vice President at Google DeepMind and Google's Chief AI Architect. The post is dated 12 November 2026 and was last updated on 13 November 2026.
Google describes Gemini 4 Carbon and Gemini 4 Argon as its frontier models for deep, long-horizon reasoning on hard problems. Hydrogen is the fast one, built for rapid execution: writing code, running tests, using tools and getting everyday work done. We covered the earlier step in Gemini 4 Carbon's opening to paying developers.
Google says it is available on all tiers of Google AI Studio and the Gemini API, in Gemini for business, in the Gemini app and in Google Search, and to paid Google AI Pro and Ultra subscribers in Google Antigravity. Google also announced Orchestrator in Antigravity, which switches between Carbon and Hydrogen inside one working session.
Google's developer guide, "What's new in Gemini 4 Hydrogen", calls it the recommended upgrade from Gemini 3.8 Flash and from earlier Gemini 3 and 2.5 models.
Key details
| Item | What Google says |
|---|---|
| Model ID | gemini-4-hydrogen, on the free and paid tiers of the Gemini API |
| Paid API price | $1.50 per million input tokens, $7.50 per million output tokens, flat across the 1 million-token context window |
| Batch price | $0.75 input, $3.75 output per million tokens |
| Gemini 3.8 Flash price | $0.75 input, $3.75 output through 31 December 2026; $1.50 and $7.50 from 1 January 2027 |
| Thinking levels | Low, medium, high (the default) and a new xhigh |
| Speed against 3.8 Flash | Tasks finished 2.5 times faster at the default setting |
| Intelligence Index score | 51 at the default level, above Argon's 50; 54 at xhigh |
| Where staff will meet it | Gemini app model menu, Google Search, Gemini for business, Gemini in Google Sheets |
All prices are in US dollars, from Google's Gemini API pricing page, and every benchmark figure here is Google's own. Google says that against 3.8 Flash, Hydrogen more than triples its score on Terminal-Bench 4.1, a coding test run at the command line (57% against 18.2%), and cuts fabricated claims by more than five times. In early testing, partners saw it run 2x to 2.8x faster while using 12% to 23% fewer tokens. Independent tests have yet to confirm them.
A token is a small piece of text, roughly part of a word, and it is how AI providers measure and bill use. A "thinking level" sets how much the model reasons before it answers: more thinking usually means better answers on hard tasks, but also more tokens and a longer wait.
Why it matters
Fast, cheaper models do most of the AI work inside business software: answering customer questions, sorting emails and running the many small steps an AI agent takes. Our guide to permissions, data access and cost for AI agents explains why each of those steps adds to the bill.
Google is betting that speed and reliability are worth paying more for. Hydrogen's standard API price is the same as the price Google already lists for Gemini 3.8 Flash from 1 January 2027, so for apps built on Flash the higher rate is coming either way. The question for a business is whether Hydrogen does enough more per task to earn it.
Alibaba went the other way with its new flagship: Qwen4-Max lists below Qwen3.8-Max and claims around 60% lower cost per task.
Reliability is the other half of the pitch. Google says Hydrogen reduces overreach in agent tasks, where a model does more than it was asked, by up to 81%, and is up to 2x less likely to get around permission limits when a task is blocked. It reports a 94% honesty rate on its internal agent test and more than 2.3x better resistance to indirect prompt injection, where hidden instructions in a web page or document try to take over the model.
Search changes too. Google says Hydrogen is upgrading Search for all users globally, handling questions with several parts or typos better, giving more visual answers and linking to a broader range of sites. Google does not say how this changes traffic to any one site, so watch your own website's enquiries rather than guess.
What this means for businesses
For most small and medium businesses, the change arrives through tools you already use. Three areas are worth checking.
Software built on Gemini. If a developer or supplier built your chatbot, quoting tool or document workflow on Gemini 3.8 Flash, the switch is more than a new name. Google's migration checklist says Hydrogen thinks at the high level by default instead of medium, rejects the minimal setting with an error, drops explicit caching (a way to reuse a long prompt more cheaply) and expects the temperature, top_p and top_k settings to be removed. Every token is billed, thinking tokens included. The same lesson came up when OpenAI set retirement dates for GPT-4o and GPT-4.1: plan model changes, do not discover them.
Staff on free accounts. Google's pricing page says data sent on the free tier is used to improve its products, and data on the paid tier is not. The OAIC recommends, as best practice, that organisations do not put personal information, particularly sensitive information, into publicly available AI tools, and it expects due diligence before you adopt a commercially available AI product (OAIC guidance on commercially available AI products).
Spreadsheets and everyday documents. Google says Gemini in Google Sheets with Hydrogen produces outputs over 30% faster than previous models for eligible business subscriptions. A person should still check any formula it writes.
A short checklist for this month:
- Ask your developer or software supplier which Gemini model your systems call, and whether they plan to move to gemini-4-hydrogen.
- Run the same real tasks on 3.8 Flash and Hydrogen at the medium and high thinking levels, and compare answer quality, speed and cost per task, not price per token.
- Keep simple, high-volume jobs such as sorting or routing on Gemini 3.5 Flash-Lite, which Google still calls its fastest, lowest-cost model.
- Budget for 1 January 2027, when the listed Flash price rises to match Hydrogen's.
- Move work that touches customer or staff details off free accounts and onto a paid business account.
When you compare several AI models, the cost of each task matters more than the headline rate. We made the same point when DeepSeek opened web search and file reading in its API. If your software connects Gemini to your CRM, accounting or ordering systems, our business systems and integrations work covers testing and switching models without breaking the connections. If you need to hire someone to own this work in-house, a recruitment agency such as Talynx can run the search. For a review of your own setup, ask us for a free first consultation and a written quote.
Frequently asked questions
What is Gemini 4 Hydrogen?
Gemini 4 Hydrogen is Google's fast, everyday model in the Gemini 4 family, announced on 12 November 2026. It is built for quick coding and agent work, while Carbon and Argon handle deep reasoning.
How much does Gemini 4 Hydrogen cost?
On the paid tier of the Gemini API it costs $1.50 per million input tokens and $7.50 per million output tokens, including thinking tokens, flat across the 1 million-token context window. Batch requests are listed at $0.75 input and $3.75 output.
Is Gemini 4 Hydrogen more expensive than Gemini 3.8 Flash?
Yes, for now. Google's pricing page lists Gemini 3.8 Flash at $0.75 input and $3.75 output per million tokens through 31 December 2026, so Hydrogen's standard rate is double that. From 1 January 2027 the listed Flash price rises to the same $1.50 and $7.50.
Can my staff use Gemini 4 Hydrogen without a developer?
Yes. Google says it is in the Gemini app's model menu, in Google Search for all users, in Gemini for business, and in Gemini in Google Sheets for eligible business subscriptions.
Does Google use what we send to Gemini 4 Hydrogen to improve its products?
Google's pricing page answers Yes for the free tier and No for the paid tier. Keep work that involves personal information on a paid account and check your privacy obligations first.
Should we switch our software from Gemini 3.8 Flash straight away?
Not without a test. Google's migration checklist lists real changes, including a new default thinking level and no explicit caching. Run your real tasks on both models and compare quality, speed and cost first.
Key takeaways
- Google released Gemini 4 Hydrogen on 12 November 2026 as its fast, everyday Gemini 4 model.
- Paid API use costs $1.50 input and $7.50 output per million tokens, double Gemini 3.8 Flash's current rate.
- Flash's listed price rises to the same level on 1 January 2027.
- Switching software needs testing first.
- Staff already meet it in the Gemini app, Search and Sheets; keep personal data off free accounts.