For Gemini 3.8 Flash, the model most projects start with, developers who read their own AI Studio pages found 5 requests per minute (RPM) and 20 requests per day (RPD) on the free tier, and 1,000 RPM and 10,000 RPD on Tier 1. Flash-Lite gives the most free room at 15 RPM and 500 RPD. These are dated readings, not a Google table: as of November 15, 2026, Google's rate-limit page prints per-model numbers for only two models and sends everyone else to AI Studio.
| Model | Free tier | Tier 1 | Where the numbers come from |
|---|---|---|---|
gemini-3.8-flash | 5 RPM · 250,000 input TPM · 20 RPD | 1,000 RPM · 2,000,000 TPM · 10,000 RPD | Free and Tier 1: read in AI Studio on one project, imisofts.com, Oct 21, 2026 |
gemini-3.6-flash, gemini-3-flash-preview | 5 RPM · 20 RPD | Shown in AI Studio | Read from 429 bodies on a new unbilled project, aiplug.work, Sep 2, 2026, and again Oct 21 |
gemini-3.5-flash-lite, gemini-3.1-flash-lite | 15 RPM · 250,000 TPM · 500 RPD | 4,000 RPM · 150,000 RPD | Free: both readings above; Tier 1: Oct 21 reading |
| Gemini 4 Carbon, Gemini 4 Argon, 3.1 Pro Preview | Not available (0) | Shown in AI Studio | Google's pricing page lists no free tier for them |
gemini-omni-1.2-flash (video) | Not supported | 8 RPM · 40 RPD | Google's rate-limit page |
| Gemini Decision 1 (Decisions API only) | 200 RPM · 50,000 TPM · 10,000 RPD | 1,500 RPM · 1,000,000 TPM · 50,000 RPD | Google's rate-limit page |
Three rules come from Google itself: limits apply per project, not per API key; RPD resets at midnight Pacific time; and "specified rate limits are not guaranteed." Tier 2 and Tier 3 numbers for the text models appear in AI Studio's tier comparison; as of November 15, 2026 no public, dated reading of them stated its source, so they are left out here.
That is also why a copied "Free = 15 RPM, Tier 1 = 300 RPM" table no longer plans anything. A usage tier is one input. The model, project, billing account, request path and current account standing also matter, and the numbers that apply to you are the ones AI Studio shows for your project today.
The practical question is therefore not only "What is the Tier 2 RPM?" It is: which project is making this request, which model and route is it using, what limits are active for that combination, and which dimension rejected the request? Once those facts are known, a 429 RESOURCE_EXHAUSTED response becomes an operational problem rather than a guessing exercise.
Gemini API free tier vs paid tier: what changes when you link billing
Linking a billing account moves a project from Free to Tier 1, and on Flash that is the difference between 20 and 10,000 requests a day. Using the readings in the table above and Google's billing and verification pages, read November 15, 2026:
- Volume. Gemini 3.8 Flash goes from 5 to 1,000 RPM (200 times) and from 20 to 10,000 RPD (500 times). Flash-Lite goes from 15 to 4,000 RPM and from 500 to 150,000 RPD.
- Models. Gemini 4 Carbon, Gemini 4 Argon and 3.1 Pro Preview have no free tier on Google's pricing page; the October 21 free-project reading showed a limit of 0 for Argon and 3.1 Pro Preview. They become callable on Tier 1.
- New limits appear. Tier 1 adds a $10 spend limit per rolling 10 minutes and a $250 monthly cap on the billing account. Free has neither.
- Verification. Google may require phone verification before some free projects can call the API (
403 FREE_TIER_VERIFICATION_REQUIRED). Paid projects do not need it. - Data terms. Google's pricing page says free-tier content is used to improve its products and paid-tier content is not.
The full free list by model, what free projects can and cannot call, and what to check when a billed project still shows free limits are in the Gemini API free tier guide.
Where the 200 RPM, 250-a-day and 15 RPM figures come from
Several numbers circulate as "the Gemini API limit." Most are real numbers attached to something other than a Flash project:
| Figure you may have seen | What it actually is | Applies to a Flash API project? |
|---|---|---|
| Free tier "200 RPM and 10,000 RPD" | The Free row for Gemini Decision 1, a classification model on the separate Decisions API. Google search summaries on November 15, 2026 presented it as the general free tier | No |
| 8 RPM and 40 RPD | Tier 1 for gemini-omni-1.2-flash video generation; Google doubled paid Omni limits on November 5, 2026 | No |
| 250 requests a day | Gemini CLI signed in with a free Gemini API key, Flash only (geminicli.com, last updated June 18, 2026) | Only inside Gemini CLI |
| 1,500 or 2,000 requests a day | Gemini CLI per user with Google AI Pro or Ultra | Only inside Gemini CLI |
| 1,500 RPD | Default limit for Grounding with Google Search on Gemini 3 and newer, per project | Only search-grounded requests |
| "15 RPM, 1 million tokens" free, "150–300 RPM" Tier 1 | Older figures from 2025 and early-2026 guides, including an earlier version of this page | No longer matches 2026 readings |
A usage tier is not a universal model quota
Why doesn't Google publish one RPM table per tier? Because the active limit depends on the model, project and account as well as the tier, and Google shows it per project instead. Google's current Gemini API rate-limit documentation says exact limits depend on several factors, including usage tier, and should be viewed in Google AI Studio. It also warns that specified rate limits are not guaranteed and actual capacity may vary.
That changes how a per-tier guide should be used:
- The tier explains which class of limits and billing cap the account can receive.
- AI Studio shows the active model limits for the project.
- Usage shows how much of each limit the project is consuming.
- The error and request route tell you which control may have rejected the call.
Do not infer a project's current capacity from an old screenshot, another account, or a different model in the same tier. Preview and experimental models are generally more restricted, and a new model revision may have different limits from the model it replaces. Limits also move without a new model: Google doubled the paid-tier limits of gemini-omni-1.2-flash on November 5, 2026, according to its release notes.
Current tier qualifications and spend controls
As of November 15, 2026, Google's rate-limit page lists these qualifications. These are billing-account rules, not promises of a particular model RPM.
| Usage tier | Current qualification | Billing-account monthly cap | Rolling 10-minute spend limit |
|---|---|---|---|
| Free | Active project or free trial; some projects need verification | Not applicable | Not applicable |
| Tier 1 | Set up and link an active billing account | $250 | $10 |
| Tier 2 | $100 paid and 3 days since the first successful payment | $2,000 | $200 |
| Tier 3 | $1,000 paid and 30 days since the first successful payment | $20,000–$100,000+ | $200 |
The monthly caps also appear in the billing guide. The rolling spend limits are evaluated over ten minutes; earlier versions of this page gave Tier 2 as $50, and Google's page now says $200. Google says whether spend-based limits apply also depends on billing history and account standing. Crossing one can return 429 RESOURCE_EXHAUSTED even when request count and input tokens appear below their active limits.
The monthly cap behaves differently from the 10-minute limit. When a billing account's total reaches its tier cap, Google pauses the Gemini API for every project linked to that account until the 1st of the next month, unless the cap is raised. Developers who hit it in October 2026 reported a 429 reading "Your project has exceeded its monthly spending cap" (one report, another). AI Studio also lets project owners set their own monthly cap per project on the Spend page; Google labels that feature experimental and warns that charges can overrun it during a roughly ten-minute processing delay.
New billing accounts are on Prepay: setup asks for at least $5 of credit, and when the balance reaches $0 requests fail with 402 Payment Required, not 429. Retrying a 402 does nothing until credit is added.
For Tier 2 and Tier 3 qualification, Google counts cumulative spending on Google Cloud services across the projects tied to the linked billing account, not only charges from one Gemini API key.
Projects linked to the same Cloud Billing account inherit its usage tier and account cap. API keys do not have independent billing settings: a key inherits its project, and all keys in that project consume the project's shared limits. Creating more keys in one project therefore does not multiply capacity.
Google automatically upgrades eligible accounts, subject to processing and review. Free to Tier 1 typically takes effect immediately; later upgrades normally apply within ten minutes. The Projects page in AI Studio is the place to confirm the tier instead of assuming a payment has already propagated. What each tier's monthly cap buys in tokens on each model, with monthly cost examples, is in Gemini API pricing and quotas.
The dimensions that can stop a request
Google describes three common dimensions:
- RPM — requests per minute. A burst can exceed RPM even when each request is small.
- Input TPM — input tokens per minute. Google counts input tokens only. Large prompts, attached context and parallel requests can exhaust input TPM before RPM.
- RPD — requests per day. This resets at midnight Pacific time, not at local midnight.
Some models add their own controls. Image-capable models may use images per minute (IPM); other models carry a tokens-per-day limit, and free-tier 429 bodies list a per-day input-token quota next to the request quotas. Built-in tools have their own limits too: Grounding with Google Search on Gemini 3 and newer defaults to 1,500 requests per day per project. Every applicable dimension is evaluated independently. Staying below RPM does not protect a request that exceeds input TPM, RPD, IPM, TPD or a spend-based limit.
A useful capacity estimate is:
requests per minute ≈ min(active RPM, floor(active input TPM / average input tokens per request))
This is an estimate, not a new quota. It helps explain why a service sending 40,000 input tokens per request can hit a token limit with far fewer calls than a service sending short prompts. With the Tier 1 Flash reading above, min(1,000, floor(2,000,000 / 40,000)) is 50 requests a minute, so tokens bind long before the 1,000 RPM figure. Add headroom for traffic variance and do not plan at 100% of a specified limit, because Google does not guarantee that capacity.
Project, model and route must match the dashboard
How do you check your own Gemini API rate limits? Open the AI Studio rate-limit page for the project that owns the key, select the exact model, and read RPM, TPM and RPD there. Before diagnosing a 429, record the key’s identifier or label, project, model ID, and endpoint used by the failing process. Keep the secret key value out of logs and notes. Then compare like with like:
- On the Projects or API keys page, confirm which project owns the key and which billing account and tier it inherits.
- In Dashboard, open the rate-limit view and select the exact model and tier used by the request. The Compare tiers dropdown on the same page shows what the other tiers would give, according to a Google forum moderator's May 2026 answer.
- Open Usage for the same project and time window.
- Check whether the call is ordinary interactive traffic, Priority inference, or Batch API traffic.
Priority inference has its own rate limits. Google currently states a default of 0.3 times the standard rate limit for each model and tier, while Priority consumption still counts toward overall interactive traffic. A dashboard value for standard requests cannot be applied unchanged to Priority traffic.
Batch API requests are separate from non-batch calls. Google currently documents 100 concurrent batch requests, a 2 GB input-file limit, 20 GB of file storage and model-specific enqueued-token limits; for Gemini 3.8 Flash those are 3 million tokens on Tier 1, 400 million on Tier 2 and 1 billion on Tier 3. The enqueued-token table changes with models and tiers, so use the live Batch limits rather than copying its entire matrix into capacity plans.

Diagnose the 429 before choosing a fix
Use the first matching cause, not every workaround at once.
| What the project shows | Likely limiting control | Action that addresses it |
|---|---|---|
| A short burst reaches active RPM | Requests per minute | Queue requests, smooth concurrency and retry after a bounded delay |
| Few requests carry very large prompts | Input TPM | Reduce repeated context, control parallelism, reuse application results where appropriate or retrieve only necessary input |
| Calls stop after sustained daily use | RPD or model-specific TPD | Wait for the documented reset or move an eligible workload to an appropriate paid tier |
| Expensive calls fail within a busy ten-minute period | Spend-based rate limit | Wait for the rolling window, reduce context/output cost, or request a higher limit if normal traffic requires it |
| Every call fails and the message mentions the monthly spending cap | Billing-account or project monthly cap | Raise the cap on the Spend page or wait for the 1st of the month; retries will not help |
The error says limit: 0 | The model has no allowance on this tier | Use a model the tier includes, or link billing for a paid-only model |
| Only Priority traffic fails | Priority inference limit | Check the Priority value; use standard traffic only if its latency is acceptable for the workload |
| Async jobs cannot be enqueued | Batch concurrent jobs or enqueued tokens | Wait for jobs to finish, reduce batch size, or check that model's live Batch table |
| Dashboard usage is low and the error persists | Wrong project/model, delayed tier state or capacity variation | Verify the key-project pair, exact model ID and tier; then submit a limit request or support case with those details |
Upgrading a tier is not the universal answer. It will not fix a process that is reading the wrong project's dashboard, using a preview model with tighter limits, or flooding the service with synchronized retries.
Read the quotaId in the 429 body
Which limit did you hit? The status line says only 429 RESOURCE_EXHAUSTED, but the JSON body names the counter. Look for a google.rpc.QuotaFailure detail and its quotaId. One free-tier response posted on November 14, 2026 listed four violations for gemini-2.5-pro at once (public log); one of them looked like this:
{
"description": "Generate content requests per minute limit exceeded on Free Tier.",
"quotaMetric": "generativelanguage.googleapis.com/generate_content_free_tier_requests",
"quotaId": "GenerateRequestsPerMinutePerProjectPerModel-FreeTier",
"quotaDimensions": { "model": "gemini-2.5-pro", "location": "global" }
}When a quotaValue field is present, it is the limit itself; the September reading in the table at the top was taken that way.
| What the 429 says | Limit it names | Clears | What to do |
|---|---|---|---|
quotaId contains GenerateRequestsPerMinute | RPM for that model | Within a minute | Back off with jitter and lower concurrency |
quotaId contains InputTokensPerModelPerMinute | Input TPM | Within a minute | Send fewer or smaller prompts per minute |
quotaId contains GenerateRequestsPerDay | RPD | Midnight Pacific | Stop retrying; switch model, project tier or wait |
quotaId contains InputTokensPerModelPerDay | Input tokens per day | Midnight Pacific | Same as RPD |
quotaId ends in -FreeTier | A free-tier counter | — | If you have paid, the key belongs to a project still on Free; follow the billed-project checks |
Message contains limit: 0 | No allowance for this model on this tier | Not on this tier | Change model or link billing |
| "Your project has exceeded its monthly spending cap" | Monthly cap | Cap raised, or the 1st | Spend page in AI Studio |
The 10-minute spend-based limit also returns 429 RESOURCE_EXHAUSTED. As of November 15, 2026 no public copy of its exact message was available, so if none of the rows above match and your traffic is expensive, compare the last ten minutes of spend with the $10 or $200 limit for your tier.
Two nearby errors are not rate limits at all. 402 means the Prepay balance is $0. 403 means a permission problem, including a free project that still needs verification. Google's troubleshooting guide says not to retry 400, 402 or 403.
Retry transient failures without creating a retry storm
Google's troubleshooting guide recommends exponential backoff for transient errors such as 429 and 503, with random jitter and a maximum attempt count. Official SDKs include retry handling: Google says the Python SDK retries 429 and 5xx errors up to four times, starting at about one second with a maximum delay of 60 seconds. Check the SDK behavior before adding another retry layer, or the two layers multiply each other.
Backoff improves recovery after a temporary window; it does not increase RPM, TPM, RPD or spend capacity. If every worker retries at the same interval, the next burst can recreate the same failure. Use one queue or shared limiter for the project rather than independent per-key limiters, because keys in the project share quota.
Do not trust the retryDelay of a per-day 429. Developers have posted per-day free-tier responses whose RetryInfo said "34s" (inspect_ai, September 2026) and "39s" (August 2026), although the counter does not reset until midnight Pacific; in the first case an evaluation kept retrying for over ten minutes until it was stopped by hand. Google's own Gemini CLI reads the quotaId instead and treats any PerDay or Daily quota, any limit: 0, and any retry delay over five minutes as final (source). Do the same: retry per-minute 429s, stop on per-day ones.
Do not retry 400 or 403 as if they were temporary rate-limit errors. They usually indicate a bad request, unsupported parameter, authentication or permission problem. The Gemini API error guide separates those cases.

Turn a dashboard snapshot into a capacity decision
For a production decision, record the date, billing account, project, tier, exact model ID, route, active RPM/TPM/RPD or model-specific limits, observed input size, concurrency and peak ten-minute spend. That small record is more useful than a universal tier table because another engineer can reproduce the comparison later.
Use the free-quota guide when the question is whether unpaid access fits a prototype. Use the billing guide when the account must qualify for another tier. Use this page when the task is to explain effective capacity and a 429 across all tiers.
The safe operating rule is simple: treat the tier as eligibility, AI Studio as the current project record, and the failing request as the unit of diagnosis. That approach remains useful even when Google changes a model, a threshold or the dashboard again.
Gemini API rate limits FAQ: free tier, Tier 1, reset time and AI Pro
What are the Gemini API free tier rate limits?
Google does not publish them for the text models. Readings of free projects on September 2 and October 21, 2026 found 5 RPM and 20 RPD on Gemini 3.8 and 3.6 Flash, and 15 RPM and 500 RPD on 3.5 and 3.1 Flash-Lite, with 250,000 input tokens per minute. Gemini 4 Carbon, Gemini 4 Argon and 3.1 Pro Preview have no free tier. Every free model is listed in the free tier guide.
What are the Gemini API Tier 1 rate limits?
An October 21, 2026 reading of AI Studio's tier comparison showed Gemini 3.8 Flash at 1,000 RPM, 2,000,000 input TPM and 10,000 RPD, and the newer Flash-Lite models at 4,000 RPM and 150,000 RPD. Tier 1 also carries a $10 limit per rolling 10 minutes and a $250 monthly cap. Check your own project's numbers on the AI Studio rate-limit page.
When do Gemini API rate limits reset?
Requests per day reset at midnight Pacific time. Per-minute limits clear within the minute, the spend-based limit within its rolling 10-minute window, and a billing-account monthly cap on the 1st of the month.
Are Gemini API rate limits per API key or per project?
Per project. All keys in a project share its limits, so a second key in the same project adds nothing. A key in a different project draws on that project's limits, while every project on one billing account shares the account's tier and monthly cap.
How do I increase my Gemini API rate limit?
Move up a tier: link billing for Tier 1, then Tier 2 follows automatically after $100 paid and 3 days, Tier 3 after $1,000 and 30 days. If normal traffic still needs more, Google's rate-limit page links a form to request a paid-tier increase, with no guarantee of approval.
Does Google AI Pro or Ultra raise Gemini API rate limits?
No. Google's AI plans page says plan benefits for developer usage apply only in the AI Studio web interface and are separate from Gemini API usage tiers. Gemini CLI does give subscribers a larger daily allowance when they sign in with the subscribed Google account. If the limit you hit is AI Studio's own chat message, see “You've Reached Your Rate Limit” in AI Studio.






