Gemini 4 Hydrogen pricing: how much the coding model costs
Gemini 4 Hydrogen pricing: $1.50 input and $7.50 output per million tokens, with a free tier. Where the bill gets pricey.
Gemini 4 Hydrogen pricing: Google's coding model for agents costs $1.50 per million input tokens and $7.50 per million output tokens on the paid tier, and it also has a free tier. Google made it available on November 12, 2026 in AI Studio and the Gemini API. But cheap per token is not cheap per task.

Gemini 4 Hydrogen pricing: what it costs and what it includes
According to Google AI Studio's documentation, Gemini 4 Hydrogen (gemini-4-hydrogen) costs on the paid tier:
| Item | Price per 1M tokens | |---|---| | Input | $1.50 | | Output | $7.50 | | Cached input | 95% less than regular input (about $0.075) |
It is available on both the free and paid tiers, with usage limits on the free one. Like Gemini 4 Carbon and Gemini 4 Argon, it has a unified one-million-token context window.
Three things Hydrogen brings
- A fourth thinking level,
xhigh, alongsidelow,mediumand the default,high. The deepest is xhigh. - A 95% discount on cached input: if you cut what repeats across your requests, that part costs far less.
- It is the default model in Antigravity, the Gemini agent in Gemini Managed Agents.
Google also says Hydrogen makes things up more than five times less than Gemini 3.8 Flash. That is a figure Google states, not a published benchmark. Google does not publish coding benchmarks for Hydrogen either, so there is no basis to say it codes better than another model.

Where the bill gets pricey
Reasoning tokens are billed as output. At $7.50 per million output tokens, a deeper thinking level can raise the bill even when the per-token price looks low. That is why the math is best done per task, not per token.
How to apply it in your business?
- Pick a real coding task you do today with another model.
- Run it on Hydrogen at one thinking level and note the input and output tokens, reasoning included.
- Work out cost per task: input × $1.50 plus output × $7.50, per million tokens.
- Use caching on what repeats in every request.
- Show the math to whoever signs off on AI spend on your team before choosing the model.
If you are comparing with another model in the family, read the full breakdown of Gemini 4 Carbon and its launch pricing.
FAQ
How much does Gemini 4 Hydrogen cost?
$1.50 per million input tokens and $7.50 per million output tokens on the paid tier. Cached input costs 95% less.
Is there a free tier?
Yes. It is available on the free and paid tiers of AI Studio and the Gemini API, with usage limits on the free one.
What is xhigh?
A fourth, deepest thinking level, added to low, medium and high (the default).
Does Hydrogen code better than other models?
Google does not publish coding benchmarks for Hydrogen. What it says is that it makes things up more than five times less than Gemini 3.8 Flash, a figure the company states itself.
Conclusion
The Gemini 4 Hydrogen pricing of $1.50 and $7.50 is attractive, but reasoning is billed as output. Instead of guessing your bill, work it out per task.