Gemini 4 Hydrogen has a clear Google AI Studio API price: $1.50 per million input tokens and $7.50 per million output tokens on Standard paid usage. Output includes thinking tokens. Google also lists a free tier and lower Batch/Flex rates, so the useful question is not just “what is Hydrogen’s price?” but “which product, tier, and task are on the bill?” The Gemini app, Antigravity, Google Cloud and third-party gateways do not share one interchangeable price. This page covers Gemini API token pricing; read our Hydrogen review for the model verdict and the Argon comparison for a side-by-side decision.
A launch-week Reddit thread asked whether Hydrogen had promotional pricing and described sticking with Flash to limit spend. That is one user’s experience, not an API rate source; Google’s rate card determines the figures here. Read the discussion.
A launch-week Reddit thread asked about promotional pricing and described staying on Flash to limit spend. That is one user’s experience, not an API rate source; the figures below come from Google’s rate card. Read the discussion.
Key takeaways
Standard paid API rates are $1.50/M input and $7.50/M output; thinking tokens are included in output.
Batch and Flex list $0.75/M input and $3.75/M output. They are separate service tiers, not a discount you can assume for every request.
Standard implicit context caching is $0.075/M cached input; explicit caching is not supported on this listing.
Google Search grounding and Maps requests may add tool charges; the API page describes their free monthly allowances and subsequent rates.
A token estimate is not a subscription quote or a cost-per-successful-task result.
Gemini 4 Hydrogen API prices at a glance
| Gemini API tier | Input / 1M | Output / 1M | Cache input / 1M | Important condition |
|---|---|---|---|---|
| Standard paid | $1.50 | $7.50 | $0.075 | Output includes thinking; implicit caching only |
| Batch paid | $0.75 | $3.75 | $0.0375 | Separate batch processing tier |
| Flex paid | $0.75 | $3.75 | $0.0375 | Separate lower-priority tier |
| Standard free | Free | Free | Free | Model access and limits apply; product data use differs |
| Priority paid | $2.70 | $13.50 | $0.135 | Higher-priority tier; separate from Standard |
Rates checked on November 16, 2026 against Google’s live Gemini API pricing page. Google says every processed token is billed on paid usage, including thinking tokens and internal tool instructions/function declarations when grounding is used. The same page marks free-tier content as used to improve Google products and paid-tier content as not used for that purpose; check the current terms that apply to your account.
Pricing timing matters for the Argon comparison. On Nov 16, Google lists Argon Standard at $2/M input and $10/M output through Nov 30, 2026, then $4/M and $20/M starting Dec 1. Hydrogen remains listed at $1.50/$7.50. The $0.40 same-token example below describes Argon only through Nov 30; from Dec 1 it would be $0.80 at those rates.
What a request costs in practice
For a simple request that uses 5,000 input tokens and returns 1,000 output tokens at Standard rates:
(5,000 × $1.50 + 1,000 × $7.50) ÷ 1,000,000 = $0.015Local calculator
Estimate Standard API costs
Use billed input and output tokens for one request. Values stay in your browser.
Paid Gemini API Standard, USD. Output includes thinking tokens. Taxes, tools and subscriptions are excluded. Google API pricing
The model page lists a unified context limit of 1,048,576 tokens, a 65,536-token default synchronous output cap, and long decode up to 1,048,576 output tokens. The calculator uses paid Standard rates and enforces the synchronous output cap.
At 100,000 input and 20,000 output tokens, the estimate is $0.30. This arithmetic excludes Search grounding, Maps, retries, and any application hosting. At the same token volume, Argon’s currently listed standard rates ($2/M input and $10/M output) yield $0.40. That is a 25% lower Hydrogen token bill for the same token counts; it does not predict which model will finish a real task with fewer retries or tokens. See the current Gemini API model list for model IDs, and the Gemini 4 Hydrogen access guide for product-specific routes.
The billing lines people miss
Thinking tokens are output tokens for billing
Google explicitly says Hydrogen’s output rate includes thinking tokens. A response that produces a short visible answer may still involve more billed output tokens internally. Record the returned usage metadata and the actual invoice rather than estimating from the visible text alone. The same caution applies to function declarations and internal system instructions when using grounding.
Caching is not a flat discount
The Standard tier lists implicit cache reads at $0.075 per million tokens and says explicit caching is unsupported. Batch and Flex show $0.0375/M cached input. Do not multiply an assumed cache percentage by every prompt: confirm that the platform reports cached tokens for your requests, and use the tier-specific price. Google’s context caching guide explains the API behavior. The official table is the source of truth for which cache options are priced.
Search and Maps can be separate line items
The API pricing page currently lists 5,000 Google Search grounding requests per month shared across Gemini 3 and newer models, then $14 per 1,000 requests, with a daily cap. Maps has a separate request allowance and rate. A user request can trigger more than one search query, so count tool executions rather than treating “one prompt” as “one tool call.” These charges sit beside model tokens, not inside the $1.50/$7.50 rates.
How to compare tiers
| Choose this route when… | Likely starting tier | Check before estimating |
|---|---|---|
| You need interactive, ordinary API responses | Standard | Standard token use, cache hits, tools, current paid/free eligibility |
| Jobs can run asynchronously | Batch | Batch support and latency tolerance |
| You can accept lower service priority | Flex | Availability, queue behavior, and task deadline |
| A request needs elevated service priority | Priority | Premium rate and whether the workload needs it |
| You are trying the model without production guarantees | Free | Model access, quotas, data-use terms, and rate limits |
Google’s developer API page has a general description of free, paid, and enterprise tiers. Those program descriptions do not make consumer subscriptions equivalent to API credits. A Gemini app plan or Antigravity usage meter is a separate product rule. For access paths, use the Hydrogen guide; for app and browser workflow distinctions, see Gemini 4 Argon browser use and Gemini 4 Carbon browser use.
Model price versus task cost
The public cost comparison that drew early attention is Artificial Analysis’s Intelligence Index task: it reported a $1.87 Hydrogen High cost per task in the November 12 release evaluation. That number is not a price quote for your codebase. It reflects a particular task set, model setting, token mix, and evaluator methodology. Google’s own model card separately reports benchmark results, while Artificial Analysis measures its own index. Treat both as evidence about those setups, not as your monthly budget. Read the AA evaluation page and methodology before repeating the figure.
A practical budget uses your own usage logs:
API token cost = input tokens × input rate / 1,000,000
+ output tokens × output rate / 1,000,000
+ cache and tool charges
cost per successful task = total API cost, including retries / successful tasksRun at least one representative week, segment by workload, and keep app quota separate from API invoices. For a browser-based research workflow, the browser is another part of the test environment; our AI browser guide and agentic browser overview explain why page state and tool use affect completion time. Browser automation is a workflow layer, not a Gemini API price plan.
Is Gemini 4 Hydrogen worth its API price?
For API users, Hydrogen is relatively easy to price because the Standard rates are visible and thinking tokens are explicitly included. Its unit rates are below Argon’s listed Standard rates, and Batch/Flex cost less still when their service trade-offs fit. But that is only the start of the decision. A model that needs more tokens, repeats, external tools, or human repair can cost more per completed job.
Use Hydrogen when its quality, speed, and actual token consumption justify the rate for your task. Keep a cheaper model for routine requests if it already meets your acceptance checks. Compare against Gemini 4 Argon and the Hydrogen model resources only after choosing the same API tier and recording the same task inputs. The Gemini 4 Carbon overview is about a different model and workflow; it does not change this API rate card.
FAQ
How much does Gemini 4 Hydrogen cost?
Google lists a standard Gemini API rate of $1.50 per million input tokens and $7.50 per million output tokens. Thinking tokens are included in the output rate; API pricing is separate from app subscriptions and managed-product quotas.
Is Gemini 4 Hydrogen free?
Google lists free-of-charge input and output tokens on the Gemini API free tier, subject to model availability and limits. The paid tier is billed at the listed token rates. Check the current pricing page before budgeting.
How do I estimate a Hydrogen API request?
Multiply input and output tokens by their respective per-million rates, divide by one million, and add separately billed tools such as Search grounding. This estimate does not apply to Gemini app or Antigravity quotas.
Does Hydrogen charge for thinking tokens?
Yes. Google states that output pricing includes thinking tokens and that every processed token is counted and billed on the paid API tier.
Is Gemini 4 Hydrogen cheaper than Argon?
On Google AI Studio standard API rates, Hydrogen is $1.50/$7.50 and Argon is $2/$10 per million input/output tokens. That makes Hydrogen 25% lower on both token rates, but it does not establish lower cost per successful task.