API pricing
One credit buys one record, not one HTTP call. A request that hands back a hundred startup profiles costs a hundred credits whether you asked for them in one page or in twenty; a request that only resolves a name to an id costs one.
Plans
- —
- 100 records per page
- No overage: stops at zero
- $0.0095 per credit
- 500 records per page
- Overage $0.015
- $0.0052 per credit
- 1,000 records per page
- Overage $0.007
- $0.0033 per credit
- 1,000 records per page
- Overage $0.004
- $0.0019 per credit
- 5,000 records per page
- Overage $0.002
Get a key — 500 free credits Talk to us about a plan
What a request costs
credits = max(1, ceil( base + records × credits_per_record × (1 + 0.25 × expands) ))
Endpoints fall into seven groups. The group decides the two numbers in that formula.
| Group | Base | Per record | What is in it |
|---|---|---|---|
| Free | 0 | 0 | Dictionaries (/v2/signals/types), this price list over the API
(/v2/pricing), and every 4xx and 5xx response. |
| Lookup | 1 | 0 | Search and autocomplete (/v2/startups/search,
/v2/funds/search, /v2/companies/search), categories and
facets, tag and location lookups, totals. Flat: the answer identifies records without
handing them over. |
| Profile | 0 | 1 | /v2/startups, /v2/funds, /v2/persons and their
{id} forms; a category resolved into one of them. |
| Relation | 0 | 0.5 | Funding rounds, post-money valuations, fund portfolios, funds raised, the people at a company, applications. |
| Feed | 0 | 0.25 | News, the /v2/signals change feed, activity, mentions, tweets, articles. |
| Enrichment | 25 / 10 | 0 | /v2/companies/find and friends crawl a site we have never seen: 25.
/venturematching runs an ML pass: 10. Neither is a database read. |
| Bulk | per dataset | /v2/dump/* and /v2/export/*. Priced at what paging through
the same rows would cost — a dump buys speed, not a discount. |
|
Worked examples
| Request | Credits |
|---|---|
GET /v2/startups/search?q=stripe | 1 |
GET /v2/startups/{id} | 1 |
GET /v2/startups?$top=100 | 100 |
GET /v2/startups?$top=100&$expand=FundingRounds | 125 |
GET /v2/startups?$top=100&$filter=... matching 3 rows | 3 |
GET /v2/news?$top=200 | 50 |
GET /v2/signals?limit=50 | 13 |
GET /v2/companies/find?website=... | 25 |
$select narrows the payload, not the price. Asking for three columns of a
startup still costs a full record. If all you need is an id and a name, the
/search endpoints do that for one credit however many they return.
Rules worth knowing
- Errors are free. A 401, 403, 404, 429 or 500 costs nothing.
- A request that matched nothing still costs 1. The query ran.
- Each
$expandadds 25% to the per-record weight, because it puts another record's worth of data on every row. - Page size never changes the price of a record. Ten pages of ten cost the same as one page of a hundred, so page at whatever size suits you.
- You are charged before the response is built, then refunded. A request that
reserves a hundred credits and returns three rows is settled back to three. The
X-Credits-Chargedheader on the response is the final figure.
Knowing where you stand
Every metered response carries:
X-RateLimit-Limit: 15000 monthly credits on your plan X-RateLimit-Remaining: 14211 credits left, after this request X-RateLimit-Reset: 1767225600 unix seconds until the grant refills X-Credits-Charged: 100 what this request actually cost
GET https://parsers.vc/api/v1/usage reports the balance and the last 30 days,
and GET https://vcapi.parsers.vc/v2/pricing returns this whole page as JSON
— plan list, group weights, and the price of every endpoint — so an agent can
budget a job before it starts. Both are free.
Running out
On Free, zero means zero: the API answers 429 with
Retry-After and the credits refill on the first of the month.
On a paid plan the monthly credits are a subscription, not a ceiling. Past them the API keeps answering and the extra records are billed by actual use at the plan's overage rate, on the next invoice. Each account has a spending cap — twice the monthly allowance unless you ask for more — so a runaway script cannot generate an unbounded bill. Your account page shows where the cap stands; to move it, write to [email protected] and we will raise it.
A refusal names the shortfall, so an agent can adapt instead of retrying:
HTTP/1.1 429 Too Many Requests
Retry-After: 1209600
{
"error": { "code": "credits_exhausted", "message": "..." },
"credits": { "limit": 15000, "remaining": 40, "required": 100, "plan": "growth",
"overage_used": 30000, "overage_cap": 30000, "reset": "2026-10-01T00:00:00Z" },
"actions": [
{ "type": "purchase_credits", "url": "https://parsers.vc/api_page/pricing/" },
{ "type": "raise_overage_cap", "contact": "[email protected]" },
{ "type": "reduce_page_size", "parameter": "$top" },
{ "type": "wait", "until": "2026-10-01T00:00:00Z" }
]
}
Whole datasets
Bulk dumps are not on self-service plans — write to [email protected] and we will unlock the datasets you need. They are priced at the same per-record weights, so a dump costs what paging through it would: the full startup set is about 913,000 credits, funds about 28,000, news about 2,448,000. Extracting the entire database comes to roughly 1.3 million credits for the core entities and around 11 million with the news and signal history. If that is what you are after, a licence is cheaper than the API and we would rather sell you one.
Questions
Why per record and not per request?
Because a request is not what costs us anything, and it is not what you came for. Charging per call means the same data is cheap or expensive depending on how you page it, which rewards writing an awkward client and penalises an honest one.
Do unused credits roll over?
The monthly allowance does not — it refills on the first of the month, UTC. Credits you buy on top of a plan sit in a separate balance and do not expire.
Can I see what a job will cost before running it?
Yes. GET /v2/pricing returns the weight of every endpoint, and
$count tells you how many records match a filter for one credit. Multiply.
What about the website subscriptions?
Those are separate. Startups Pro and VC firms Pro unlock the site; these plans are for the API.
Full API reference · Using the API through an assistant · Instructions for the agent itself · Your keys and usage