Google Places API Limits in 2026: Complete Reference

Anyone who has ever opened the Google Cloud console, enabled the Places API, written ten lines of code, and started a loop has run into the same wall within minutes. The requests stop. The error codes appear. The dashboard fills with red. Understanding the full set of Google Places API limits in 2026 is not optional if you want to ship anything that touches Google Maps data, because every single endpoint is gated by quotas, rate limits, radius caps, pagination caps, field-level pricing, and a free tier that disappears faster than most people expect. This guide is the consolidated reference: what every limit actually means, where the hidden ones live, and when it makes more sense to skip the API entirely.

Quota types: per-day, per-second, per-method

The Places API does not have a single quota. It has a layered system. Every project gets a per-second quota that controls burst behavior, a per-day quota that controls volume, and per-method quotas that scope those numbers to specific endpoints such as Nearby Search, Text Search, Place Details, Place Photos, and Autocomplete. Hitting any one of them returns the same generic OVER_QUERY_LIMIT response, which is what makes debugging painful. You see the symptom, but not which of the three layers tripped.

The practical consequence is that you cannot just plan for "ten thousand requests per day." You also have to plan for the second-by-second shape of those requests, and you have to plan for which methods you are calling, because the per-method ceilings do not stack the way developers expect. A high Place Details quota does not buy you a higher Nearby Search quota.

Rate limit per second (QPS)

The Places API enforces a queries-per-second cap at the project level. The default sits in the low hundreds of QPS for most active endpoints, with separate counters for each method. Bursting above the cap returns an immediate error, even if your daily quota is nearly empty. Google does grant quota increases on request for production projects with billing history, but the review process is manual and not instant.

For most builders, the QPS limit is the first one they hit. A naive scraper that fires requests in parallel will saturate the per-second cap long before any daily ceiling. The fix is client-side throttling, exponential backoff on retry, and respecting the Retry-After hint when it appears. None of this is exotic engineering, but all of it is engineering you have to write, test, and maintain. For a wider tour of the surrounding ceilings, the Google Maps API limits explained guide covers the broader Maps Platform picture.

Daily and monthly quota

The Places API does not bill on a daily quota in the traditional sense. Billing is monthly, calculated against the SKUs you trigger, and a $200 free credit is applied each month before charges begin. The daily quota you see in the console is a safety cap you set yourself to prevent runaway spending, not a hard product limit. That distinction matters: you can raise the daily quota toward the platform ceiling, but every request above the free credit threshold becomes a real invoice line.

In practice, teams set the daily quota deliberately low for the first weeks of a project, then raise it once the cost-per-request math is understood. Forgetting to set it is one of the most common ways a side project ends up with a four-figure bill on day one.

Pagination limits: 60 results maximum

This is the hidden cap that breaks more lead-gen workflows than any other. Nearby Search and Text Search return at most 20 results per page, and at most 3 pages per query, for a hard ceiling of 60 results regardless of how many businesses actually exist in the area you searched. There is no parameter to raise it. There is no pro tier that unlocks more. Sixty is the number.

If you are searching for "restaurants in Paris," there are tens of thousands of restaurants in Paris, and the API will hand you 60 of them. The standard workaround is to subdivide the search area into smaller geographic tiles, run a separate query against each tile, deduplicate the overlapping results by place_id, and stitch the output back together. This is doable, but the cost in requests, code, and edge-case handling is substantial.

next_page_token delay

Pagination has a second subtle limit. When the API returns a next_page_token, that token is not immediately valid. Calling it within the first second or two reliably returns INVALID_REQUEST. The required delay floats between roughly two and five seconds depending on server-side propagation, and there is no documented exact value. Builders are expected to retry with backoff until the token activates.

This means a full three-page search is not three round-trips. It is three round-trips with two enforced waits between them, so a single Text Search query takes between five and fifteen seconds end to end before you have all 60 results. Multiply by the number of geographic tiles you are subdividing, and the wall-clock time adds up quickly.

Radius limits: 50,000 meters

Nearby Search accepts a radius parameter measured in meters, and the maximum accepted value is 50,000. Anything larger is silently capped. Combined with the 60-result cap, this means a single Nearby Search query can never cover a large metropolitan area exhaustively. You will always be subdividing.

Text Search behaves slightly differently. It does not require a radius and can interpret broader location strings, but it still returns at most 60 results, so the practical exhaustiveness ceiling is the same.

Field-level pricing: Basic, Contact, Atmosphere

The Places API does not charge a flat rate per request. It charges per SKU, and the SKU is determined by which fields you ask for in the response. Asking only for place_id, name, geometry, and a handful of other low-tier fields keeps you in the Basic Data SKU. Asking for phone numbers, website, opening hours pulls you into the Contact Data SKU at a meaningfully higher per-request rate. Asking for ratings, reviews, price level, or user content pulls you into the Atmosphere Data SKU at the highest tier.

The dangerous default is to request all available fields "just in case." Doing that on every Place Details call multiplies your bill by roughly three to five times compared to a careful field selection. The Google Maps API pricing vs scraping breakdown walks through the SKU math in more detail.

Field masks: how to lower cost

Field masks are the official mechanism for telling the API exactly which fields you want and paying only for those tiers. A request with a tightly scoped field mask of name, place_id, formatted_address, and geometry stays Basic. The same request without a field mask, or with a wildcard, walks into Contact and Atmosphere automatically.

The discipline is straightforward but unforgiving: every endpoint that supports field masks should use them, every code path should be reviewed, and every new feature that needs an additional field should be evaluated against the SKU it triggers before going to production. Forgetting once on a hot path is how teams discover surprise invoices.

Pricing tiers and the $200 credit

Google Maps Platform applies a $200 monthly credit to each billing account, and the credit is consumed across all Maps Platform SKUs together, not per-API. A project that uses Maps JavaScript, Geocoding, and Places shares one pool. The credit covers a meaningful amount of light usage, but at production volumes it is exhausted in hours, not days.

After the credit is gone, every request is invoiced at its SKU rate, which for Places ranges from a few dollars per thousand at the Basic tier to high double-digit dollars per thousand at Atmosphere. There is no flat-rate enterprise tier published publicly; high-volume customers negotiate directly with Google sales.

Common errors and fixes

The three errors every Places API developer learns to recognize are OVER_QUERY_LIMIT, ZERO_RESULTS, and INVALID_REQUEST. OVER_QUERY_LIMIT means one of the quota layers tripped, and the fix is throttling plus backoff plus checking which specific quota in the console actually moved. ZERO_RESULTS is not an error in the strict sense, but a signal that the query plus location plus radius combination found nothing, and the fix is usually broadening the query or the area. INVALID_REQUEST during pagination almost always means the next_page_token has not yet activated, and the fix is waiting longer before retrying.

A fourth error, REQUEST_DENIED, is the one that ends evenings: it means the API key is misconfigured, the API is not enabled on the project, billing is not active, or the referrer or IP restriction rejected the call. The remedy is in the console, not in the code.

When the API is the right choice

The Places API is the right tool when you are building a product that displays Google data inside a Google-owned surface, when you need real-time freshness on a small number of places, when the fields you need are clearly inside the Basic SKU, and when your volume comfortably fits inside the $200 free credit or your unit economics absorb the per-request cost. Store locators, single-business widgets, and lightweight autocomplete experiences all fit this shape.

When MapsLeads is the right choice

If your goal is a list of leads, not a product feature, the API stops making sense. You do not want to write tile-subdivision logic. You do not want to memorize which fields are Basic and which are Contact. You do not want to build a two-to-five-second sleep into your pagination loop. You want a list of businesses with emails and phone numbers, exported to CSV, and you want it now.

MapsLeads handles the entire flow. Open the Search panel, enter a query and a city, and run the search. Results come back already deduplicated, with no 60-result ceiling to fight, no radius cap to subdivide around, and no rate limit to babysit. Each base result costs 1 credit. To unlock verified emails and direct phone numbers, enable Contact Pro for an extra 1 credit per result. Reputation data, including detailed review signals, costs 1 additional credit. Photos cost 2. Group results into named lists, dedupe across searches, and export to CSV when you are ready.

The credit model is the entire pricing surface. There are no SKUs, no field masks, no quota tickets to file with support, and no surprise invoice on the first of the month. Compare the two paths side by side in MapsLeads vs Google Places API direct, or jump straight to Pricing to see the per-credit math for your volume.

FAQ

What is the Google Places API daily limit? There is no single fixed daily limit. The console exposes a daily cap that you set yourself as a spend safety control, and the platform enforces per-method quotas plus a project-level QPS cap. Production projects can request increases.

What is the Places API rate limit per second? The default per-second quota sits in the low hundreds of QPS per method at the project level in 2026, with the exact figure depending on the endpoint. Hitting it returns OVER_QUERY_LIMIT immediately.

Does the Places API have a free tier? Google Maps Platform provides a $200 monthly credit applied across all Maps Platform SKUs combined. Light usage stays free; production volume exhausts the credit quickly.

What is the maximum number of results per Places API search? Sixty. Nearby Search and Text Search return up to 20 results per page across at most 3 pages, regardless of how many businesses actually match.

What is the Nearby Search radius limit? 50,000 meters. Larger values are capped silently.

What is the best Google Places API alternative for lead generation? A purpose-built tool like MapsLeads removes the 60-result ceiling, the radius cap, the field-mask discipline, and the QPS babysitting, and replaces them with a flat per-credit price.

Conclusion

The Google Places API limits in 2026 are not arbitrary. Each one exists for a real reason inside Google's infrastructure. But every one of them is friction if your goal is a list of leads instead of a product feature. The QPS cap, the 60-result pagination ceiling, the next_page_token delay, the 50,000-meter radius, the SKU pricing, and the $200 credit together turn what feels like a one-afternoon project into a multi-week engineering effort with an unpredictable monthly bill at the end of it.

If that is the project you actually want to build, by all means build it. If it is not, skip the entire stack. Get started with MapsLeads, run your first search in under a minute, and export a clean list of businesses with verified contact data the same day.