ApyHub logo
Api Tools · Agents Orchestration Frameworks

ApyHub

ApyHub is a Dutch API marketplace that bundles more than 1,500 certified HTTP endpoints — OCR, file conversion, data validation, geocoding, LLM calls — behind a single key and subscription, metered in a usage unit the platform calls atoms.

Active GDPR compliant Free plan Freemium API available 16+ Verified by Guidaio
Overview

What is ApyHub?

ApyHub is an API marketplace built on an unusual premise: you subscribe to the catalog, not to individual APIs. One apy-token opens more than 1,500 endpoints spread across 400-plus services, and the allowance you buy is spendable anywhere in that catalog rather than locked to the one API you bought it for. The company frames this against the incumbents it names on its own about page, RapidAPI and APILayer, where a per-API subscription means capacity idles on one endpoint while another hits its ceiling.

Usage is metered in atoms, a deliberately small unit. A cached lookup might cost one atom, a frontier-model generation up to 250, and the publisher sets the figure at the moment it answers the call. The stated aim is that a 200-byte request and a 12 MB OCR job should not cost the same, which request counting cannot express.

The catalog itself is broad and uneven. The largest sections are Artificial Intelligence (89 APIs), Developer Tools and File Conversion (52 each), Data Extraction and Geolocation (38 each) and Data Validation (31), alongside smaller sets for audio, SEO, HR, finance, security and image work. Much of it is genuinely utilitarian — currency dictionaries, JWT decoders, QR generators — and a sizable share of the geolocation coverage is United States only.

The second pitch is trust rather than breadth. Every service is reviewed before it ships and carries machine-readable certification stating where data lives, how long it is kept, which third parties touch it and which standards it claims to meet. ApyHub argues this collapses vendor auditing into one contract, and leans on recent npm supply-chain compromises to make the case.

Every endpoint is also MCP-ready: a streamable MCP server lets Claude, Cursor or any agent runtime discover and call the catalog directly. On the supply side, publishing is self-service and AI-assisted, taking roughly ten minutes, with the author keeping full IP, setting the atom price and owing no exclusivity. ApyHub B.V. operates from the Netherlands and claims 65,000-plus developer workspaces a month.

What it does

  • Call more than 1,500 certified HTTP APIs with one apy-token, across 400+ services in a single catalog
  • Convert documents, images, audio and archives: PDF, Word, Excel, PowerPoint, WebP, MP3 and dozens more formats
  • Extract text and structured data from PDFs, HTML pages, invoices and résumés, including OCR on scanned images
  • Validate regulatory and financial data: EU VAT numbers, IBAN, BIN, US routing numbers, IFSC codes, postcodes
  • Run language-model calls, from chat completions with tools and structured output to summarization, sentiment and entity extraction
  • Process images at scale: upscale, restore, deblur, remove backgrounds, watermark, blur faces, crop and resize
  • Expose the whole subscribed catalog to an AI agent as MCP tools, with no wrapper code to write
Audience

When to use ApyHub / When not to

A quick filter to help you decide if ApyHub is the right fit.

When to use ApyHub

  • Back-end and full-stack developers who would rather call a hardened endpoint than maintain another utility library
  • Product teams assembling features quickly — document conversion, OCR, address validation, geocoding — without a separate vendor per capability
  • Builders of AI agents: every endpoint is MCP-ready, so Claude, Cursor or any agent runtime can call it without a wrapper
  • Startup founders and CTOs validating an idea on the permanent free tier, 150 calls a month with no payment card
  • API authors who want to monetize an endpoint: self-service publishing in about ten minutes, full IP retained, no exclusivity

When not to use ApyHub

  • Non-developers: there is no no-code interface, every capability is reached through an HTTP call and an apy-token header
  • Teams that need a signed customer DPA or a downloadable SOC 2 report, since the site publishes neither
  • Organizations requiring data residency outside the EU on a self-serve plan, as US residency is an Enterprise-only option
  • Mobile-first users: ApyHub ships no iOS or Android application and no browser extension
  • Resellers, because the terms explicitly forbid re-exposing an ApyHub endpoint verbatim as your own product
Get started

How to use ApyHub

A typical end-to-end flow, from setup to results.

  1. Create a free account, with no payment card, or sign in through Google, GitHub or Microsoft
  2. Browse the catalog, filtering by category or by GDPR, SOC 2 and HIPAA tags, and sorting by popularity, newness or name
  3. Open a service page to read its description, version, endpoint count, publisher and the atom cost of a call
  4. Test the endpoint straight from the browser with the Try It Live block, which prefills your default key
  5. Mint an apy-token in one click; the same key authenticates every API in the catalog
  6. Copy the ready-made snippet in cURL, Node, Python, Go or Java, or generate a typed client locally
  7. Pass the key in the apy-token header of your HTTP request and read the JSON response
  8. For agent work, point your runtime at the MCP server and let Claude or Cursor discover the tools; MCP is enabled from Pro+ upward
  9. Track atom consumption in the dashboard, where history runs 7 days on the free tier and 30 to 90 days on paid plans
  10. Prepurchase atoms or move up a tier as the quota approaches; once it is exhausted traffic stops rather than overbilling you
Quick read

Pros & Cons

Pros

  • One contract, one key and one invoice for 1,500+ APIs, which collapses the vendor auditing surface
  • Pooled allowance: atoms bought against one API are spendable on any other, unlike per-API marketplace plans
  • Prices are public and low at the entry point, USD 16 a month, with no sales call and no minimum commitment
  • A permanent free tier with no payment card gives full catalog access from day one
  • EU hosting by default in Frankfurt with Dublin backups, five named subprocessors and 30 days' notice before changes
  • An explicit commitment not to train models on customer data and not to sell it
  • Traffic stops when the quota runs out, so there is no surprise overage bill

Cons

  • The pricing page renders in JavaScript and shows no amounts without a browser; the figures come from the public plans endpoint that feeds it
  • Atoms do not translate cleanly into calls — a request costs 1 to 250 atoms depending on the endpoint — which makes cost forecasting hard
  • SOC 2 and ISO 27001 are claimed on the homepage and per endpoint as alignment, with no trust page, report or certificate published
  • No customer-facing DPA is offered; the only agreements mentioned are those ApyHub signs with its own subprocessors
  • No Enterprise plan appears in the public grid, so US data residency and custom terms require a quote
  • Support leads with a Discord community, with email presented as the secondary route and no published response SLA
  • Catalog quality is uneven: several sections hold only two APIs, many are thin utilities, and much of the geolocation coverage is US-only
Pricing

Pricing & Plans

A permanent free plan is available. Starter costs nothing and requires no payment card, providing 1,000 atoms a month capped at five calls a day, which is the 150 monthly calls advertised on the homepage. The cheapest paid tier is Pro at USD 16 per month (EUR 15), raising the allowance to 100,000 atoms. Taxes are added where required, and no time-limited free trial is offered.

Starter — USD 0 / EUR 0 per month
  • 1
  • 000 atoms
  • 5 requests per second
  • 5 calls per day
  • 50 MB per file
  • 50 MB storage
  • 1 API key
  • 1 user
Pro+ — USD 42 / EUR 40 per month
  • 500
  • 000 atoms
  • 10 requests per second
  • 300 MB per file
  • 10 GB storage
  • 10 API keys
  • 5 users
  • 90-day history
Team — USD 86 / EUR 80 per month
  • 2
  • 000
  • 000 atoms
  • 15 requests per second
  • 10 users
  • 20 API keys
  • MCP enabled
Scale — USD 270 / EUR 250 per month
  • 6
  • 000
  • 000 atoms
  • 50 requests per second
  • 20 users
  • 40 API keys
  • MCP enabled
Plan 6
  • Starter
  • Pro and Pro+ are presented as the Medium Traffic range
  • Team and Scale as Heavy Traffic
Enterprise — not listed in the public grid
  • referenced in the terms and privacy policy through a signed order form
  • invoice billing and an optional US data-residency region
Plan 8
  • Prepurchased atoms are added to the monthly quota
  • once the quota is exhausted traffic stops and no further billing occurs
Plan 9
  • Self-serve plans renew until cancelled
  • with cancellation effective at period end
  • no minimum commitment
  • and no refund of unused credits on downgrade
Prices and plans listed above may evolve. Always check the official pricing page before subscribing.
Trust & Privacy

Data, GDPR & hosting

A consolidated view of how ApyHub handles your data.

GDPR overview

GDPR implementation is unusually concrete. The privacy policy, version 3.2 effective 18 April 2026, enumerates all seven data-subject rights and wires each into the product: a JSON export for access, profile settings for rectification, a danger zone for erasure, notification settings for objection. The lead supervisory authority for EU users is named, the Dutch Autoriteit Persoonsgegevens, a Data Protection Officer answers at dpo@apyhub.com, and replies are promised within two business days. Five subprocessors are listed by name and region, each under a signed Data Processing Agreement, with 30 days' notice before any change. Analytics cookies are gated behind consent. No Article 27 representative is designated, and none is required: the controller is established in the Netherlands. The homepage badge reads SOC 2 and GDPR, but no trust page or certificate is published.

Who owns the data?

Section 04 of the terms casts ApyHub as a conduit that takes no copyright in what passes through it. Request bodies are forwarded to the underlying API provider, and ApyHub keeps only the metadata needed to bill the call. Response bodies belong to the customer, but under the license of the specific API invoked: some grant full ownership, others license the output for narrow uses only, and each license is surfaced on the catalog page before use. Authors who publish an API keep ownership of it and take 85% of the revenue it earns. Account, workspace, usage and payment data sit with five named subprocessors, and ApyHub states it never sells any of it.

Reuse rights

Whether an output can be reused depends on the license attached to the endpoint that produced it, not on a single marketplace rule, so the answer changes from one API to the next and the terms point customers to the catalog page before they build on a result. Building products on top of ApyHub is expressly permitted; reselling raw API access, or re-exposing an endpoint verbatim as your own product, is not. Section 03 also rules out illegal content, denial-of-service traffic, scraping behind authentication, credential stuffing, malware, spam, phishing and infringing material. Enforcement suspends first and investigates second, with a reversal and a credit if ApyHub got it wrong. Usage data serves billing and debugging only, and marketing email is opt-in with the default off.

Data retention & training

Retention summary
Retention is set per category. Account data lives as long as the workspace and is deleted within 30 days of its removal. Usage logs are kept 90 days then purged automatically, down from 180 days in June 2025, while aggregate monthly totals are held longer for billing records. Invoices and receipts are kept seven years for tax compliance and are the only category that cannot be deleted on request. Support tickets are kept two years after closure, then anonymized. Backups run on a rolling 30-day window, so a deletion takes up to 30 days to propagate fully. Request and response bodies are never logged, only call metadata such as endpoint, timestamp, status code, latency and workspace ID, and logs are redacted before any human reads them. Customers can export everything as JSON or delete the workspace themselves from the settings.
Trains on customer data
No
Subprocessors disclosed
Yes
GDPR contact

Hosting summary

Primary data is stored on AWS in eu-central-1 (Frankfurt, Germany), with backups replicated to eu-west-1 (Dublin, Ireland) on a rolling 30-day window. Edge caching runs globally through Cloudflare, where cached data lives for seconds to minutes and, ApyHub states, never contains personal information. Signing up from the United States does not change the location: data is stored in the EU by default, and US residency is offered only as an Enterprise option, the stated reason being that GDPR protects everyone regardless of where they signed up. Five subprocessors are named with their regions: AWS in the EU, Cloudflare globally, Stripe across the US and EU, Postmark in the US and a self-hosted Sentry in the EU. Card details never reach ApyHub's servers, as Stripe holds them and ApyHub sees only a customer ID and the last four digits. The domain resolves to a Cloudflare anycast node geolocated in the United States, which is a CDN point of presence and not a storage location. Data is disclosed on legally binding request, with prior notice to the customer unless prohibited.

Hosting countries
🇩🇩 Germany🇮🇪 Ireland
Hosting regions
EU
Watch-outs

Things to keep in mind

Risks and trade-offs to weigh before adopting ApyHub.

  • Single-intermediary dependency: one gateway outage touches every API at once, and the uptime target explicitly excludes the upstream APIs themselves
  • The publisher sets the atom cost when it answers and may raise it without renegotiation, so cost per call can climb after you integrate
  • Traffic halts the moment the quota is exhausted, which can take a production service down through nothing worse than a busy week
  • Certification is asserted without a verifiable document, so trust rests on the vendor's own statement rather than an audit you can read
  • Request bodies are forwarded to a third-party provider, so sending personal or confidential data means first reading that specific API's license
  • Output ownership varies by API; assuming one uniform rule across the catalog is a genuine legal risk
  • The habit of never writing a utility function again erodes in-house skill and makes an exit expensive if a provider disappears or reprices
Setup

Setup & Integrations

Technical difficulty

Low for anyone comfortable with HTTP. Account creation needs no payment card and accepts Google, GitHub or Microsoft sign-in; a single apy-token then goes in a request header, with no OAuth flow and no request signing. Endpoints can be tested in the browser before writing code, and snippets ship in cURL, Node, Python, Go and Java, with a first call advertised inside 60 seconds. The real prerequisite is issuing an HTTP request and reading JSON, since there is no no-code interface. The harder task is not technical: estimating how many atoms a workload will consume.

Deployment

Web appAPI

Integrations

Claude Cursor Stripe Google GitHub Microsoft

Supported languages

English
Company

Behind ApyHub

Company name
ApyHub B.V.
Founded
26/11/2021
Country of origin
🇳🇱 Netherlands
Headquarters
Eindhoven, Netherlands
UBO
INFORMATION_NOT_FOUND
UBO country
INFORMATION_NOT_FOUND
Domain registrar country
🇺🇸 United States
Legal contact
Support contact

Social

Official links

Resources

All the official URLs gathered for verification and reference.

Compare

Alternatives

Tools that compete with or complement ApyHub.

R RapidAPIA APILayer
FAQ

Frequently asked questions

What is an atom on ApyHub?
It is the platform's billing unit. A cached lookup can cost a single atom while a frontier-model generation can cost 250, and the publisher declares the figure at the moment it answers the call. That is how a 200-byte request and a 12 MB OCR job end up priced differently.
Is there a free plan, and does it need a card?
Yes. The Starter plan is permanently free and requires no payment card. It includes 1,000 atoms a month, capped at five calls a day, which is the 150 monthly calls quoted on the homepage. There is no separate time-limited free trial.
How much does the cheapest paid plan cost?
Pro costs USD 16 a month, or EUR 15, and raises the allowance to 100,000 atoms with 10 API keys and 5 users. Above it sit Pro+ at USD 42, Team at USD 86 and Scale at USD 270 a month.
Do I need a separate key for each API?
No. One apy-token authenticates every API in the catalog, passed in the apy-token header. There is no per-vendor signup, no separate account and no approval step before you can call an endpoint.
Can an AI agent call ApyHub endpoints?
Yes. Every endpoint is MCP-ready by default and ApyHub runs a streamable MCP server, so Claude, Cursor or any agent runtime can discover and invoke the catalog without wrapper code. MCP is enabled on the Pro+ tier and above.
Where is my data stored?
Primary data sits on AWS in eu-central-1 (Frankfurt, Germany), with backups replicated to eu-west-1 (Dublin, Ireland). US signups are still stored in the EU by default; US residency exists only as an Enterprise option.
Is my data used to train AI models?
ApyHub states it does not use customer data to train models, build profiles, serve ads or sell insights, and that its revenue comes from subscriptions alone. Request and response bodies are not logged at all, only call metadata.
What happens if I exhaust my monthly quota?
Traffic stops. ApyHub states it does not bill beyond the quota, so there is no overage invoice. You either prepurchase additional atoms, available from the Pro plan upward, or move to a higher tier.
What uptime is committed, and is there a minimum age?
A 99.99% monthly uptime target applies to the platform gateway on Pro and above, with service credits from 10% to 50% of the monthly fee if it is missed, claimable within 30 days. Users must be at least 16 years old.
How do I get support, and can I publish my own API?
Support runs primarily through the Discord community, with hello@apyhub.com, live chat and an on-page AI assistant as alternatives. Publishing is self-service and AI-assisted, takes about ten minutes, splits revenue 85/15 in the author's favour and requires no exclusivity.
Conclusion

Should you pick ApyHub?

ApyHub answers a real and specific complaint: that buying APIs one subscription at a time leaves teams paying for headroom on one endpoint while rationing another. Subscribing to the catalog instead, with a pooled allowance and a billing unit scaled to the work each call actually does, is a coherent fix rather than a repackaging, and the entry price is low enough that testing the claim costs almost nothing.

The governance is stronger than the category average. EU hosting by default, five subprocessors named with their regions, 30 days' notice before any change, retention periods stated category by category, a reachable Data Protection Officer and an explicit refusal to train on customer data are all published rather than implied. Traffic stopping at the quota, instead of generating an overage invoice, is a genuinely customer-favouring default.

Two gaps deserve attention before anyone commits production traffic. SOC 2 and ISO 27001 appear as a homepage badge and as per-endpoint alignment, with no trust page, audit report or certificate to inspect — and alignment is not certification. There is likewise no customer-facing DPA, which will stall any procurement review that requires one. The pricing page also depends on JavaScript to show its own figures, which is an odd weak point for a product that sells reliability.

The catalog is broad but uneven: alongside solid OCR, conversion and validation work sit thin utilities, two-entry sections and geolocation coverage largely confined to the United States. Atoms bring fairness at the cost of predictability, since a call can range from 1 to 250 of them. This is a developer product, and without HTTP integration skills there is nothing here to use. For teams that fit, it is a credible way to stop maintaining the boring middle of their own stack.