Gemini 4 Hydrogen: Faster AI, Price Change Jan 1
Google launched Gemini 4 Hydrogen on Nov 12, 2026 at $1.50 per million input tokens, and Gemini 3.8 Flash doubles in price on Jan 1. What to check.
By Peter Walters
Google announced Gemini 4 Hydrogen on November 12, 2026, a new model that replaces the "Flash" name and is priced at $1.50 per million input tokens and $7.50 per million output tokens. Google says it finishes tasks 2.5 times faster than Gemini 3.8 Flash, and Gemini 3.8 Flash itself doubles to the same price on January 1, 2027. If a chatbot or automation of yours runs on a Gemini Flash model through the API, that date matters more than the launch.
What actually changed
Hydrogen is Google's new everyday model. According to 9to5Google's launch report, it takes a 1 million token input, is the default model for AI Mode in Google Search, and reaches free Gemini app users through Auto model selection, where the picker labels it "Everyday help." Google calls it a major standalone upgrade with significant leaps in speed and intelligence.
Developer pricing is $1.50 per million input tokens and $7.50 per million output tokens, flat across the full context window. Beri.net's breakdown says the Gemini API changelog lists it as generally available on November 12, with a free tier alongside the paid one.
Here is the part that is easy to miss. Google's own pricing page shows Gemini 3.8 Flash at $0.75 input and $3.75 output today, rising to $1.50 and $7.50 on January 1, 2027. Batch, Flex, Priority and caching rates double on the same date. After that, the old model and the new model cost the same per token.
On cost per finished task, Beri.net cites Artificial Analysis figures of $1.87 per task for Hydrogen at its default setting versus $7.77 for Gemini 3.8 Flash. The same source reports Intelligence Index scores of 51 and 31. Those are third-party benchmark numbers on someone else's tasks, so treat them as a reason to test, not a promise.
Why it matters if you run a small team
Most small businesses will never type a model name. You meet this change in two places: the free Gemini app, which just got better quietly, and any tool a developer built for you that calls the Gemini API.
The second one is where money moves. If someone built you a website chatbot, an inbox sorter or a booking assistant on Gemini 3.8 Flash, your bill per message is scheduled to double in about six weeks, and nobody will email you about it. Hydrogen may bring it back down if it finishes tasks in fewer tokens. Partners quoted in the 9to5Google report saw 12% to 23% lower token usage. That is a vendor-side claim, so check it against your own traffic.
What is overhyped: the benchmark race. A score of 51 versus 50 against Google's own Argon model tells a florist or a yoga studio nothing. What is not overhyped: a model that is fast and cheap enough that a chatbot answers in a blink, because slow replies are why most website bots get ignored.
One real caution. Beri.net reports Hydrogen refuses more multimodal prompts than Flash did. If your tool reads photos, receipts or screenshots, test that before you switch anything.
What I would do with it this week
You do not need to learn the model. You need to ask three questions.
- Ask whoever built your AI tools which Gemini model they call, and what you pay per month for it.
- Ask them to run Hydrogen against 20 real messages or documents from your business and compare quality, speed and cost.
- Put January 1 in the calendar so the price change does not surprise you.
If you use the free Gemini app, just try it on a real task: summarizing a long email thread, drafting a client reply, cleaning up a spreadsheet. Judge it on whether it saved you time.
When I build a chatbot or an assistant for a client, I keep the model name in one setting rather than scattered through the code. Swapping to a cheaper or better model is then an afternoon of testing, not a rebuild. If your tool does not work that way, now is a good time to ask why. And I would test on real examples first, never on a vendor demo.
Sources
- Google announces Gemini 4 Hydrogen, 9to5Google
- Gemini API pricing, Google AI Studio
- Gemini 4 Hydrogen launch, pricing and cost per task, Beri.net
- What is Gemini 4 Hydrogen, Kie.ai (release date cross-check)
I write about this kind of thing every day, and you can get it in the letter.