API & MCP server
Every figure on this site is available as JSON and as an MCP tool. No key, no sign-up and no CORS restrictions to start — a key only raises the ceiling, it never changes the answer.
62 tools, one per question the site answers, each returning a
discriminated union (ok: true or
ok: false with a machine-readable code) and an
attribution block naming the dataset, the provider, the
reference year and a ready-to-quote citation sentence.
Try it without signing up
curl "https://geo-parity.com/api/calculate.json?income=60000&target=VN"
That works right now, from anywhere, with no key. The same call is the MCP tool
calculate_ppp — MCP requests dispatch in-process to these same handlers, so an
agent and a curl get byte-identical payloads.
Connect the MCP server
Add this to your client's config (Claude Desktop, Claude Code, or anything else that speaks MCP). The package is a thin stdio bridge to the hosted server — there is nothing to keep updated when tools are added.
{
"mcpServers": {
"geo-parity": {
"command": "npx",
"args": ["-y", "geo-parity-mcp"]
}
}
}
Or point a remote MCP client straight at https://geo-parity.com/api/mcp. The server is open —
no auth — and metered by IP exactly like the REST API. It negotiates protocol revisions
rather than asserting one, and serves tools, resources and prompts.
Limits, and what a key buys
There are two ceilings: a monthly quota, which is the allowance a plan buys, and a per-minute burst, so one caller cannot saturate the service. Quotas reset on the 1st, UTC. Metering is per account, not per key — holding several keys does not raise the allowance.
| Plan | Price | Requests/month | Burst/minute |
|---|---|---|---|
| Anonymous no key, no account | Free | 1,500 per IP | 30 |
| Free account a key, so the quota follows you | Free | 2,000 per account | 60 |
| Lite the AI adviser, and a little more API | £1/mo | 4,000 per account | 60 |
| Pro saved comparisons and CSV export too | £4/mo | 20,000 per account | 90 |
| Pro Plus more of everything Pro has | £12/mo | 75,000 per account | 150 |
| API Starter for a product in the background; Pro Plus features included | £19/mo | 150,000 per account | 200 |
| API Scale high-volume or agent fleets; the largest adviser allowance | £59/mo | 600,000 per account | 300 |
Prices are in GBP and exclude any tax your country applies; USD pricing and checkout are on the pricing page. Billing runs through Polar, which is the merchant of record.
Create a key on your account page — it takes a minute and the free tier needs no card.
Handling the ceilings
Every response carries X-RateLimit-Limit, -Remaining and
-Reset for the minute, and X-Quota-Limit, -Remaining
and -Reset for the month. Two different 429s, and they mean different things:
-
RATE_LIMITED— too fast. Wait forRetry-Afterand continue; your monthly allowance is untouched. -
QUOTA_EXCEEDED— the month's allowance is spent. Retrying before the 1st cannot help; a key or a higher plan is the only thing that does.
Send a key as Authorization: Bearer gp_live_…, or ?key=… where you
cannot set headers.
Specifications
- openapi.json — OpenAPI 3.1 for every endpoint
- llms.txt — the capability index, written for models
- .well-known/mcp.json — the MCP server card
- .well-known/api-catalog — RFC 9727 linkset
Every indexable page also has a Markdown twin at index.md beside it, so an agent
can read the prose without parsing HTML.
What stays free
The API stays open and key-free. Paid plans raise the ceiling; they do not put a dataset, an endpoint or a tool behind a login, and nothing here is priced per call. If the free tier covers what you are doing, that is the intended outcome — the plans exist for volume, not to ration the answers.