AiStaffo

Google releases Gemini 4 Hydrogen with a free API tier

Stock photo: laptop desk documents

Google released Gemini 4 Hydrogen on 12 November 2026. On the Gemini API it costs $1.50 per million input tokens and $7.50 per million output tokens. Google presents it as a fast model for everyday coding and agent tasks, and it is the first Gemini 4 model with a free API tier, which carries rate limits. Hydrogen is priced above the Gemini 3.8 Flash model it succeeds, so the cost case depends on the work. For automation teams, the practical question is which steps need a paid model and which can run within the free allowance. The Gemini 4 Carbon and Argon introductory prices of $2 and $10 are scheduled to rise to $4 and $20.

In short

  • Gemini 4 Hydrogen launched on 12 November 2026 at $1.50 per million input tokens and $7.50 per million output tokens.
  • It is the first Gemini 4 model with a free API tier, which carries rate limits.
  • Hydrogen is priced above the Gemini 3.8 Flash model it replaces, so it is not a cost cut for existing Flash workflows.
  • Gemini 4 Carbon and Argon introductory rates are scheduled to double, so budgets should use the later rates.
  • Test cost per completed task on your own documents before changing any production workflow.

What happened

Google released Gemini 4 Hydrogen on 12 November 2026. According to Google's launch post, the model is built for quick execution of agent workflows and everyday tasks, and it is available across the Gemini API, Google AI Studio, Gemini for business, the Gemini app and Google Search. Google also brings Gemini 4 to all tiers of Google AI Studio and the Gemini API, which is the detail that matters most for developers. Google describes the model as its daily driver for agentic work, with the larger Gemini 4 Carbon and Argon kept for the hardest reasoning tasks.

VentureBeat reported the release on the same day. Google says Hydrogen completes tasks 2.5 times as fast as Gemini 3.8 Flash at its default reasoning setting. Google Cloud lists gemini-4-hydrogen as generally available, and Google says eligible business subscriptions also get it in Gemini in Sheets.

Why it matters for businesses that automate work

Most automation work is repetitive. A model that reads an invoice, drafts a payment reminder or checks a reconciliation runs many times a day, so the price per token adds up quickly. Hydrogen is the first Gemini 4 model that can be tried without a paid plan, which lets a team test a workflow before it commits budget.

The price is not a bargain against the previous generation. VentureBeat describes Hydrogen as a faster, smarter and pricier successor to the Flash model. Notebookcheck also notes that Hydrogen replaces Gemini 3.8 Flash as Google's everyday model. A team that switches an existing workflow from Flash to Hydrogen should expect a higher bill for the same volume of tokens, unless the task finishes with fewer steps.

What changes in practice

The published rates are per million tokens. A rough example helps a non-technical reader see the scale. Suppose a back-office workflow sends 10 million input tokens and receives 2 million output tokens in a month. At Hydrogen's paid rates, that is about $30 for the month, calculated as 10 times $1.50 plus 2 times $7.50. The same volume on the introductory Carbon or Argon rates would cost more, because those models are priced higher.

ModelInput per million tokensOutput per million tokensFree API tier
Gemini 4 Hydrogen$1.50$7.50Yes, with rate limits (according to a third-party guide)
Gemini 4 Carbon, introductory$2$10Not stated in the launch post
Gemini 4 Carbon, after introductory period$4$20Not stated in the launch post

For a team that already uses an API connection, the change is a model name in the configuration, plus testing. Output quality, instruction following and the number of retries should be measured on the team's own documents before the switch. Google's launch post says Hydrogen follows instructions more closely and makes fewer fabricated claims, but a business should check those claims against its own files rather than accept them on the vendor's word.

For a team that uses the consumer Gemini app rather than the API, the change is smaller. Notebookcheck reports that free users cannot pick Hydrogen themselves. Google says free users reach it through automatic model selection, and with auto model selection turned off, every reply comes from Flash-Lite. A staff member therefore cannot rely on Hydrogen being used for a given prompt in the free app.

Limits and risks

The free tier has limits on requests per minute and per day, and those limits change over time. A workflow that runs at volume will need a paid billing account. A free tier is also not a guarantee of future access, so a business should not build a permanent process on it without checking Google's pricing page at the time of build.

Automation also cannot fix a poor process. A model that reads an invoice faster does not know whether the purchase order was approved. People still need to decide which outputs are accepted, which are sent back, and who owns an error. Hydrogen's speed and price do not change that.

What to watch next

Three points will matter over the next few weeks. The first is the date when the Carbon and Argon introductory rates end. The Google post for Carbon gives the later price of $4 per 1M input tokens and $20 per 1M output tokens, but the post does not give a date. Third-party trackers such as AIHubMix list the change as 1 December 2026, and businesses should confirm the date on Google's own pricing page before they set a budget.

The second point is the Hydrogen rate itself. Google's launch coverage notes that Google lists no introductory promotion for Hydrogen's standard tier, so the current rate is the only published rate. The third is independent testing. Artificial Analysis has already scored Hydrogen on its own index, and teams should look for results that measure cost per completed task, not only cost per token.

How AiStaffo would automate this

AiStaffo would connect the model to the systems where the work already sits, such as the mailbox, the accounting ledger or the document folder, and set Hydrogen to read incoming items and draft the outputs in a fixed format. Routine steps like classifying a supplier bill or drafting a payment reminder run automatically, and each run is logged with the token count so the cost per task is visible. The business owner or finance lead still approves anything that posts money, sends a message to a customer or changes a record. Before any switch, the free audit measures the current volume and the token cost of each step, so the decision rests on the business's own figures. Book a free automation audit.

Questions people ask

How much does Gemini 4 Hydrogen cost?
According to Google's launch coverage, Gemini 4 Hydrogen is priced at $1.50 per million input tokens and $7.50 per million output tokens on the paid API. Cost depends on how many tokens each task uses, so the total bill should be measured on real workloads. Check Google's pricing page before you budget.
Is Gemini 4 Hydrogen free to use?
A third-party guide reports that the Gemini API and Google AI Studio offer a free, rate-limited tier for Hydrogen, the first Gemini 4 model to have one. The free tier has request limits per minute and per day. Free users of the Gemini app cannot choose Hydrogen themselves and receive it through automatic model selection.
Is Hydrogen cheaper than Gemini 3.8 Flash?
No. VentureBeat describes Hydrogen as a pricier successor to the Flash model, so existing workflows may cost more after the switch. Its speed may reduce the number of steps for some tasks, which is why teams should compare cost per completed task rather than cost per token.
When do Gemini 4 Carbon and Argon prices change?
Google's Carbon launch post lists $2 and $10 per million tokens as the introductory price and $4 and $20 after the introductory period, without giving a date. Third-party trackers list 1 December 2026 as the change date. Confirm the date on Google's pricing page before setting a budget.
Does Hydrogen replace a staff member's review of automated work?
No. A model can read, classify and draft, but a person still decides which outputs are accepted, who owns an error and whether a payment or customer message is sent. Automation reduces repetitive typing and checking, not the need for judgement.

Book a free automation audit

Thirty minutes. We look at one process you run every week and tell you exactly what an AI worker would take off your desk, and what it would not.

We use these details only to reply to this request and prepare your audit, and keep them for up to two years. We send no marketing unless you tick the box above. Privacy policy

ai modelsai agentsai toolsgemini