Some links on this page are affiliate links: if you buy through them we may earn a commission, at no extra cost to you.
Gemini API cost and capacity are separate things: token and tool prices determine what usage costs, while RPM, TPM, RPD, and spend-based limits determine how much a project can use and when requests may be refused. Google’s figures vary by model, service tier, and effective date. For example, Gemini 4 Argon Standard is listed at $2 per million input tokens and $10 per million output tokens through November 30, 2026, with higher prices taking effect December 1.
To avoid surprises, check the exact model row and your project’s current limits, estimate input, output, caching, and tool usage separately, and monitor both billing and quota. A 429 error can mean a rate limit or a spend-based limit has been reached; upgrading a tier or adding an API key does not necessarily fix the underlying cause.
As an Amazon Associate I earn from qualifying purchases.
What do Gemini API pricing, quotas, and rate limits each mean?
Pricing answers how Google charges for API usage. Quotas and rate limits answer how much a project can use within specified periods. Billing-tier requirements and spending safeguards are separate controls: they affect account eligibility, capacity, and whether usage can continue.
- Pricing: model- and service-specific charges, including input tokens, output tokens, cached tokens or cache storage, and certain tools.
- Quotas: usage allowances measured in dimensions such as requests per minute (RPM), tokens per minute (TPM), and requests per day (RPD).
- Tier: the project’s access and capacity level, linked to billing status and, at higher tiers, payment history.
- Spend safeguards: limits that can stop requests when a rolling spend rate is exceeded, even if a per-minute or daily quota is not exhausted.
These controls do not substitute for one another. A project can be within its billing-account cap but hit an RPM limit, or have room under its RPM limit but receive a refusal after reaching a spend-based limit.
How much does the Gemini API cost?
There is no single Gemini API price. Google’s live pricing table lists charges by model and may separate input, output, caching, tools, and service tiers. The example below is specifically for Gemini 4 Argon Standard; it is not a rate for every Gemini model or service.
| Gemini 4 Argon Standard usage | Through November 30, 2026 | From December 1, 2026 |
|---|---|---|
| Input tokens | $2.00 per 1 million tokens | $4.00 per 1 million tokens |
| Output tokens | $10.00 per 1 million tokens | $20.00 per 1 million tokens |
Those effective dates matter: as of October 25, 2026, the listed through-November price is the current one for this example, and the later rate is scheduled to begin December 1. Check the live pricing table before budgeting or deployment because model names, rates, and effective dates can change.
The Tool Desk
Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Which usage should I estimate?
Do not multiply one headline token rate by all activity and treat the result as a complete bill. Estimate each applicable category separately:
- Input tokens: the content sent to the model.
- Output tokens: the content generated. Where a model’s pricing row says thinking tokens are billed as output, include them in the output estimate.
- Context caching: check both cached-token pricing and any storage charge listed for the model or service.
- Tools and grounding: look for separate pricing rows; tool use may add charges beyond model tokens.
- Service tier: compare Standard, Batch, Flex, or Priority prices where offered. Do not assume a model offers every tier or that their rates match.
Context window is also a model specification, not a price. A larger context window can affect how much input a request can contain, but the price still depends on the applicable billing categories and actual usage.
Can Batch API reduce the bill?
Google advertises a 50% cost reduction for Batch API on Paid tiers. That is a service-specific offer, not a universal half-price rate for every model, tool, or usage category. Verify the model’s Batch row and eligibility in the current pricing table before using the reduction in a forecast.
What are Gemini API RPM, TPM, and RPD limits?
Google generally measures rate limits as requests per minute (RPM), tokens per minute (TPM), and requests per day (RPD). The applicable values depend on the model and tier, so there is no single limit that applies to every Gemini API project.
- RPM limits the number of requests a project can send per minute.
- TPM limits the number of tokens processed per minute. A project can hit this limit even if it has not used all of its allowed requests.
- RPD limits requests over a day. Google says RPD quotas reset at midnight Pacific time.
Rate limits are applied per project, not per API key, according to Google’s Rate limits documentation. Creating additional keys for the same project therefore does not create separate project capacity. Limits also vary by model; experimental and preview models can have tighter limits, so treat the model’s status as part of capacity planning.
Google directs developers to AI Studio to see active limits, which vary by model and tier. Check the specific model and project there rather than relying on a number from another project, an older guide, or a different model.
How do Gemini API billing tiers change access and capacity?
The tiers differ in more than rate limits. Free offers limited model access and free input and output tokens. Paid adds higher rate limits, context caching, Batch API access, and advanced models. Enterprise adds support, security and compliance features, provisioned throughput, and volume discounts.
| Tier | Access and capacity described by Google | Billing qualification | Listed billing-account cap |
|---|---|---|---|
| Free | Limited model access; free input and output tokens | Free tier | |
| Paid | Higher rate limits, context caching, Batch API access, and advanced models; Google says prompts and responses are not used to improve its products | Requires Cloud Billing; $5 minimum prepaid credit to upgrade | Tier 1: $250 |
| Paid | Paid-tier features | Tier 2 requires $100 paid and three days from the first successful payment | Tier 2: $2,000 |
| Paid | Paid-tier features | Tier 3 requires $1,000 paid and 30 days from the first successful payment | Tier 3: $20,000–$100,000+ |
| Enterprise | Support, security and compliance, provisioned throughput, and volume discounts |
The billing qualifications and caps above are Google’s listed tier details; the cap is not the same as a quota for a particular model or a rolling spend limit. The $5 prepaid-credit minimum is the stated minimum to upgrade to Paid, not a promise that the credit covers a particular amount of API work. Google describes the transition as moving from free use to prepaid and pay-as-you-go pricing.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Google says prompts and responses on Paid are not used to improve its products. That privacy statement is specifically about the Paid tier; do not infer that the same terms apply to Free or Enterprise from this comparison alone. For Enterprise requirements, confirm the terms and features for the specific agreement.
What is a spend-based limit, and why can it cause a 429?
A spend-based rate limit is a rolling safeguard distinct from a token or request quota and from the billing-account cap. Google lists these 10-minute spend limits for Paid tiers:
| Paid tier | Rolling 10-minute spend limit | Listed billing-account cap |
|---|---|---|
| Tier 1 | $10 | $250 |
| Tier 2 | $50 | $2,000 |
| Tier 3 | $200 | $20,000–$100,000+ |
Google says that hitting a spend-based rate limit returns 429 RESOURCE_EXHAUSTED. The 10-minute figure is a rolling spend-rate safeguard, not a daily budget and not a per-request price. A billing-account cap is another control and should not be read as the amount you can spend in every 10-minute window.
Why am I getting 429 RESOURCE_EXHAUSTED?
A 429 response means the request was refused because a relevant limit or resource constraint was reached; it does not by itself identify which one. Check the active project and model limits and distinguish a short-window rate problem from an exhausted daily allowance or a spend safeguard.
Quick wins for a faster PC:
Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Repair Windows errors before they cause bigger problemsFix Now →- Confirm the project and model. Check that the request is using the intended project, and inspect the active limits for that exact model in AI Studio.
- Check which dimension is exhausted. Compare request volume with RPM, token volume with TPM, and daily requests with RPD. Remember that RPD resets at midnight Pacific time.
- Check spend controls. If the project has reached its tier’s rolling 10-minute spend limit, Google returns 429 even if request or token throughput appears available.
- Reduce or pace traffic. For per-minute pressure, reduce concurrent or burst traffic and retry with exponential backoff for transient rate errors. Retrying immediately in a tight loop can add load without resolving the limit.
- Reassess tier eligibility. If the workload needs more sustained capacity, review whether the project qualifies for a higher tier and whether its model is preview or experimental.
Do not assume that every 429 clears after a short wait: an RPD limit has a daily reset, while a rolling spend limit is measured over 10 minutes. Identify the applicable constraint before choosing a recovery action.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.How do I increase my Gemini API quota?
First verify which project, model, and quota dimension is limiting you. If higher capacity requires a paid tier, Google says upgrading increases rate limits and requires setting up Cloud Billing; further tier increases depend on the published payment and elapsed-time qualifications.
- Inspect the current limit: open Google AI Studio and review limits for the project and model you are calling.
- Set up billing for Paid: activate Cloud Billing and meet the stated $5 minimum prepaid-credit requirement to upgrade.
- Meet higher-tier qualifications: Tier 2 requires $100 paid plus three days from the first successful payment; Tier 3 requires $1,000 paid plus 30 days from that payment.
- Recheck capacity: verify the model’s current RPM, TPM, and RPD in AI Studio after a tier change. A higher tier does not make all models share one universal quota.
- Plan against safeguards: track the tier’s spend-based limit and billing-account cap independently from request and token quotas.
If usage is blocked by a preview or experimental model’s tighter limits, changing to a higher tier may not resolve every model-specific constraint. Check the model’s current availability and limits before designing around it.
How should I budget and monitor a Gemini API project?
A reliable estimate starts with the workload, not a single per-token headline. Separate interactive requests from jobs that can use Batch, identify any tools or grounding, and forecast the model’s input and output separately.
Recommended Free Tools
- Select the exact model and service tier. Find that exact row in Google’s live pricing table and note its effective date.
- Estimate usage by category. Forecast input tokens, output tokens (including thinking tokens when the row specifies them), cached usage and storage, and tool charges separately.
- Check context and eligibility. Verify the context window, whether the model is available on Free or Paid, and whether the chosen service tier is supported.
- Check project limits in AI Studio. Record RPM, TPM, and RPD for the project and model instead of treating another project’s values as applicable.
- Monitor both spend and activity. Watch prepaid balance, project and billing-account spend caps, and the Dashboard > Usage view.
- Revisit the assumptions. Pricing, model names, and limits are date-sensitive. Recheck the live table and project limits before launch and when updating forecasts.
For production systems, build backoff into handling transient rate errors and make retry behavior sensitive to the failure type. A daily quota exhaustion is not the same problem as a burst of requests, and neither is the same as reaching a rolling spend limit.
Frequently Asked Questions
Do Gemini API quotas apply separately to each API key?
No. Google applies rate limits per project, so multiple keys under one project do not provide independent RPM, TPM, or RPD allowances. If separate workloads need isolated usage tracking or limits, review the project structure rather than rotating keys.
When does the Gemini API daily request quota reset?
Google states that RPD quotas reset at midnight Pacific time. If your application serves users across time zones, use that reset boundary when interpreting a daily quota error.
Are Gemini API prompts used to train Google’s products?
Google says prompts and responses on the Paid tier are not used to improve its products. That statement is tier-specific; check the applicable terms for other tiers and agreements.
PC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchCan I use Gemini API for free in production?
The Free tier includes free input and output tokens, but Google describes it as having limited model access and lower limits than Paid. Whether it fits a production workload depends on its model eligibility, throughput needs, and applicable terms.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
