Gemini 4 Hydrogen: Google's Model Card and Price, November 12, 2026
Updated November 12, 2026 · first published November 12, 2026
On November 12, 2026, Google published a model card for Gemini 4 Hydrogen (model ID gemini-4-hydrogen), and its Gemini API pricing page lists it at $1.50 per million input tokens and $7.50 per million output tokens on the standard paid tier, with no dated price change. Unlike the Krypton rumours, this model has a vendor page and a price. Neither page says whether it is in preview.
What Google has published
- Model card. Google DeepMind's Gemini 4 Hydrogen card is marked "Published 12 November 2026" and says the model is adapted from Gemini 4 Carbon. It positions Hydrogen as a faster model for everyday coding and agentic workflows, while Carbon and Argon are the frontier models. It lists a combined input and output context window of up to 1M tokens, up to 64K default synchronous output tokens, and up to 1M through background long decode.
- Where it is offered. The card lists the Gemini app, Gemini API, Google AI Studio, Gemini for business, Google Antigravity, Google Sheets and Google Search.
- Price. The Google AI Studio pricing page lists the standard paid tier at $1.50 input and $7.50 output per million tokens, output including thinking tokens, and $0.075 for cached input. The note says caching is implicit only; explicit caching is not supported. The page shows no future-dated change for Hydrogen and no preview label.
Price against the other Gemini 4 models
| Model | Input per 1M, standard | Output per 1M, standard | Dated change |
|---|---|---|---|
| Gemini 4 Hydrogen | $1.50 | $7.50 | None listed |
| Gemini 4 Carbon | $2.00 | $10.00 | $4.00 and $20.00 from December 1, 2026 |
| Gemini 4 Argon | $2.00 | $10.00 | $4.00 and $20.00 from December 1, 2026 |
Against today's Carbon and Argon rates, Hydrogen is 25% cheaper per input and output token ($1.50 against $2.00, $7.50 against $10.00). Against the December 1 rates it is 62.5% cheaper ($1.50 against $4.00, $7.50 against $20.00). These are our calculations from the listed prices, not Google's. A lower token price is not a lower cost per task if a model needs more tokens to finish the work.
What the card says about quality
The card includes a benchmark table covering agentic coding, knowledge work, computer use, multimodal understanding, science and math, and reasoning, and compares Hydrogen with Claude Sonnet 5.5 and GPT-6.1 Terra. Those are Google's own results. We have not reproduced them and do not quote scores here.
What this means for a budget
- Price it on the published rate, not the codename. Hydrogen has a model ID and a rate, so a forecast can carry it as a line item with its source date.
- Count thinking tokens. The pricing page lists output as including thinking tokens, so a model that reasons at length can cost more per task than its per-token price suggests.
- Test your own workload. A cost comparison needs your own prompts, token counts per task and the same grading criteria across Hydrogen, Carbon and Argon.
What we have not verified
- Release status. The card and pricing page do not say whether Hydrogen is preview, general availability or limited access.
- Benchmarks. All scores come from Google's card; we did not run them.
- Batch, Flex and Priority rates, free-tier limits, the knowledge cutoff and any safety report. We did not check them.
- Any link between Hydrogen and the Krypton tester posts. We found none.
All figures are as of November 12, 2026. Check the pricing page before you put a number in a finance document. For the models in the December 1 price step, see the Argon December 1 increase and Gemini 4 Carbon against Claude Fable 5.2.
Frequently asked questions
Is Gemini 4 Hydrogen available?
Google published a model card for it on November 12, 2026 and lists it on the Gemini API pricing page under the ID gemini-4-hydrogen. Neither page says whether it is in preview or generally available, so check AI Studio before relying on it.
How much does Gemini 4 Hydrogen cost?
The Google AI Studio pricing page lists $1.50 per million input tokens and $7.50 per million output tokens on the standard paid tier, with thinking tokens included in output and $0.075 for cached input. No dated price change is listed.
Is Gemini 4 Hydrogen the same as Krypton?
We found no evidence that it is. Krypton has been reported only through tester posts and Google has published nothing about it. Hydrogen has an official model card and a listed price.
Are the Gemini 4 Hydrogen benchmarks independent?
No. The benchmark table is in Google's own model card, compared with Claude Sonnet 5.5 and GPT-6.1 Terra. We have not reproduced the scores.
Related
- AI API price calendar: Q4 2026
- Gemini 4 Carbon vs Claude Fable 5.2
- Gemini 4 Argon price doubles on December 1
- Krypton, Grok 4.8 and Qwen 4: what is real
Want this applied to your stack? Bring the provider bills, gateway logs, and top workflows; we will map the cost drivers and savings path. Book a free audit →