Pricing
From testing to production, pay for what you need.
Choose MCP for direct use, or API plans to integrate into your product.
Developers testing an MCP integration
Start with Google sign-in and test the core tools without an API key.
Individuals using MCP regularly
For asking and browsing yourself. To serve it inside an app or service, use a developer plan below.
Teams wiring up their first F&B feature
Capacity for wiring it up and getting it right.
- Up to 1,000 results per search
- 10,000 distinct venues / 30 days
- Priority 30 rpm
Production teams with real users
Sized for apps calling us about a thousand times a day.
- Up to 5,000 results per search
- 40,000 distinct venues / 30 days
- Priority 60 rpm
Nationwide services with heavy traffic
Sized for services calling us thousands of times a day.
- Up to 5,000 results per search
- 100,000 distinct venues / 30 days
- Priority 120 rpm
Companies needing scale and custom terms
Usage, data coverage, and support customized to your product.
- Unlimited extraction
- Bulk export · data license
- SLA · dedicated support
What do the three developer limits mean?
These three limits apply only to the developer (API) plans you build into a product. Separate from tokens (depth/usage), they set how wide, how fast, and across how many venues your live queries go. Guardrails, not nickel-and-diming. (The personal Free/Plus plans have none of these; they run on a daily token budget only.)
Up to N per search · result window
How many results a single search can page through. Starter 1,000 · Pro and Scale 5,000. Pulling a whole region at once is Bulk Export territory.
N distinct venues / 30 days · extraction cap
How many distinct venues you actually touch over a rolling 30 days. Starter 10,000 · Pro 40,000 · Scale 100,000. Serving the same popular spot to many users, again and again, still counts as one, so real recommendation apps rarely hit it; only harvesting the dataset wholesale does (then it's Bulk / a data license).
Priority N rpm · requests per minute
How many requests per minute you can send. Higher means your calls go through first, with no waiting or 429s under load. Starter 30 · Pro 60 · Scale 120.
How tokens are spent
Tiers aren't plans. They only set how many tokens a single call spends, and every plan can call every tier. Basic data spends 1, rich data spends 4, and owner-uploaded photos (images, the richest data) spend 8. page_size (how many you get per call) has no effect on cost.
| Tier | tokens / call | Fields included |
|---|---|---|
| core | 1 | id, name, category, categoryLabel, location, address, regionCode, confidence, sourceCount, possiblyClosed, lastSeenAt |
| contact_hours | 2 | phone, homepage, bookingUrl, yumiReservable, reservationChannels, snsLinks, businessHours, is24h |
| quality | 2 | reviewCount, latestReviewDate, freshnessScore, dataQuality, price, menuSummary, openNow, closureSignal, promotion, photoCount, nearestStation |
| edge | 4 | accessHints, recommendedFor, atmosphere, menu, serviceAttributes, dietary, atmosphereScores, aiSummary, reputation, influencerFeatures |
| media | 8 | photos (owner-uploaded gallery, menu & menu-board image URLs. Video on the roadmap) |
Why the live limits, and why bulk is separate
The live API is for answering queries in the moment. So storing, copying, or redistributing responses in bulk is off-limits, the same as map APIs, to stop data we worked to verify from being resold as-is. When you need to load data into a product or own a dataset, we offer it properly through Bulk Export and a data license: rights already cleared, none of the terms-and-copyright risk that comes with scraping someone else's map. That legal path simply doesn't exist on map APIs.
Talk to us · Bulk & data licensing →