OpenRegistry vs OpenCorporates
OpenCorporates has been the default open-company-data resource for over a decade. Most teams who land here are happy with what it does and stuck on one specific thing it does not: the data is not live, the free tier runs out fast, the API was not designed for AI agents, or the country they care about is thin. This page lays out where OpenRegistry takes a different approach, and where OpenCorporates is still the right pick.
Side-by-side
| OpenRegistry | OpenCorporates | |
|---|---|---|
| Coverage breadth | 27 national registries | ~145 jurisdictions |
| Coverage depth | Profile, filings, officers, shareholders, UBO, charges where the upstream registry exposes them | Name + jurisdiction + identifier for most rows; deeper fields uneven |
| Data freshness | Live; every call hits upstream at request time | Scrape and cache; varies by source |
| Field shape | Raw upstream payload plus a unified envelope | Normalised to OpenCorporates' own schema |
| Filing PDFs / iXBRL bytes | Returned inline via fetch_document |
Metadata only; bytes paywalled at source |
| Free tier | 30 req/min/user | 50/day and 200/month on the public API key |
| Free-tier scope | Commercial use permitted | Open-data / public-benefit projects |
| Cross-border ownership chain | One MCP prompt across up to 30 jurisdictions | Manual ID stitching across OC's normalised schema |
| Transport | MCP (JSON-RPC over Streamable HTTP) and REST | REST |
| Identifier integrity | Native registry identifier preserved; source_url reconstructable |
OpenCorporates ID is primary; the round-trip back is sometimes lossy |
The short version
Want the widest possible catalogue and you are happy with periodic crawls? OpenCorporates still wins on breadth. 145 is a lot more than 27.
Want a registry-grade source for an AI agent or a compliance pipeline, with live data, raw fields, filing bytes returned inline, identifiers that round-trip back to the government's own portal, and a free tier you can use commercially? That is the job OpenRegistry was built for. The depth-versus-breadth trade is intentional.
Where the depth difference shows up
- UBO chain walking. One MCP prompt walks a UK company up four corporate layers and back down to the eight humans who hold the chain. Worked example on Iceland Foods.
- Filing bytes. Annual accounts, CS01s, statutory deeds, scanned filings.
fetch_documentreturns the raw iXBRL, PDF, or XML. OpenCorporates surfaces filing metadata; the bytes typically still need a trip to the upstream paid portal. - Status semantics. "active - proposal to strike off" is a different KYB risk to plain "active". The full upstream
company_status_detailsurvives injurisdiction_data; aggregator schemas tend to flatten this kind of nuance. - Niche-deep registries. Korea OpenDART XBRL filings, Spain BORME Sección II structured publications, Italy via the EU BRIS gateway, Hong Kong via the data.cr.gov.hk Open Data API, Cayman dual-tier (CIMA plus the General Registry). Country pages below.
By topic
OpenCorporates REST against OpenRegistry MCP plus REST. Free-tier limits, signup model, 30-second integration.
OC's 50/day free cap against OpenRegistry's 30 req/min free tier. What "free for open data" actually buys you.
Live calls against scrape-and-cache. Where the lag bites and where it does not.
Statutory PSC, post-CJEU restricted registers, and how each gets handled.
By country
Pick the jurisdiction you care about. Each page covers what the upstream registry publishes, what OpenCorporates surfaces today, and what OpenRegistry adds on top.
OpenDART filings, financials, shareholders. Live XBRL bytes through fetch_document.
GCIS official register, full officer plus shareholder coverage, cross-company tracing.
data.cr.gov.hk Open Data API, free, no key, BRN equal lookup.
BORME Sección II structured XML plus a delta stream for officers and shareholders from Sección I province PDFs.
Direct EU BRIS gateway. Real-time Codice Fiscale lookup. No SPID required.
Just want to look up a UK company without evaluating tools? /companies/uk has the search box plus deep pages on director appointment tracing, PSC beneficial ownership, and filing downloads.
Frequently asked questions
How is OpenRegistry different from OpenCorporates?
OpenCorporates is a scrape-and-cache aggregator covering ~145 jurisdictions with a normalised schema. OpenRegistry queries 27 government registries live on every call, returns the registry's own payload, and ships an MCP transport so AI clients (Claude Desktop, Cursor, Cline, Goose, Zed) wire in without a custom HTTP layer. Pick OpenCorporates for breadth and bulk export. Pick OpenRegistry when you need live data, raw filing bytes, or an AI-agent integration.
Is OpenRegistry really free? What is the catch?
Yes — the free tier is free and permits commercial use. Thirty requests per minute per user across all 27 registries and all tools. Higher throughput needs Pro at $9 a month or Max at $29. There is no hidden cap on which fields are returned.
The one exception is Cayman Islands. The KY surface is paid-tier only because the upstream CIMA register costs us per query. Free callers receive HTTP 402 with an upgrade pointer.
What countries does OpenRegistry cover?
27 national registries with full or partial coverage: UK, Germany, France, Spain, Italy, Norway, Netherlands, Belgium, Czechia, Poland, Switzerland, Liechtenstein, Monaco, Cyprus, Iceland, Ireland, Finland, Isle of Man, Korea, Taiwan, Hong Kong, Cayman Islands, Australia, New Zealand, Canada (federal + BC + NT), Mexico, Russia, India, Indonesia, plus US states (NY, CA, FL). The live capability matrix is at list_jurisdictions over MCP, mirrored at /docs/.
Do I need an API key?
No. Paste the MCP URL https://openregistry.sophymarine.com/mcp into Claude Desktop, Cursor, Cline, Goose, or Zed. OAuth 2.1 with Dynamic Client Registration (RFC 7591) auto-registers the client on first call and the user signs in via passwordless email magic link.
Can I use the free tier in a commercial product?
Yes. Unlike OpenCorporates' free API, which is restricted to open-data and public-benefit projects, OpenRegistry's free tier permits commercial use without a licence conversation. The rate limit is what scales to a paid tier, not the use case.
How fresh is the data?
Every tool call is a real-time HTTP request to the upstream government registry at the moment your client asks. A short performance cache (measured in minutes) sits in front to absorb hot-page bursts, and fresh=true bypasses it. There is no daily, weekly, or monthly crawl. The data is as current as the registry's own record at the second of query.
Is the source code open?
The public docs and 10 Claude Agent Skills are CC-BY-4.0 on GitHub. The hosted MCP service is operated by Sophymarine. Self-hosting at scale is not supported by the free tier; the docker stdio bridge in the docs repo is for development only.
About this page. OpenRegistry is published by Sophymarine, so this comparison is written by a party with a commercial interest in one of the tools on it. Figures and product claims for the other vendors come from each vendor's own public material, linked inline where applicable. Reviewed 2026-04-29. Corrections welcome at hello@sophymarine.com.