upAPI vs Apify
One meter instead of seven, and a price you can read before the call.
How Apify charges
Four subscriptions — $0, $19, $199 and $999 a month — layered over pay-as-you-go compute at $0.20 per compute unit falling to $0.13 at volume, plus separate meters for proxy traffic, storage and, on rented actors, a monthly rental fee on top of usage.
How upAPI charges
One subscription. Every operation declares a weighted unit cost on its catalog page — 1 for a read, 6 for a Maps search, 20 for a browser render — and a month is one pooled budget those costs draw from. Proxies are inside the unit price.
USD 29 / mo · USD 0.29 per 1,000 units · full pricing
Last checked 2026-09-14. Apify figures here were read from their own pages on that date. We do not restate a competitor’s number we have not seen ourselves, and where they publish none we say so; if this date has aged, check theirs before ours.
Apify’s own documentation concedes the shape of the problem: users report confusion about rental pricing because they are paying a monthly rental fee and platform usage at the same time. That is not a criticism of the platform so much as of the unit — a compute unit measures how long a container ran, which is a fact about the infrastructure rather than about the thing you asked for. When a run costs more than you expected, there is nothing on the invoice you could have predicted beforehand.
upAPI moves the number to the other side of the call. Each operation publishes its weight in the catalog and in `/openapi.json`, the gateway returns your remaining balance in a response header on every request, and a hard monthly quota with a 429 is the ceiling rather than a spend alert that arrives after the money is gone. That is a smaller product than Apify. It is a much easier one to budget.
The honest summary: if you need to run your own crawler for six hours and keep the dataset, use Apify. If you need to call the same twenty data sources from an app or an agent and have the bill be the same every month, the pooled quota is the better fit.
Where Apify wins
- Breadth. Apify’s store carries tens of thousands of actors; if the thing you need is obscure, someone has probably already written it and we probably have not.
- It is a runtime, not just an API. You can push your own container, run a long crawl for hours, and keep the dataset — upAPI operations are atomic calls with a 28-second edge cap.
- Datasets, key-value stores and scheduling are first-class. We deliberately keep orchestration out of the product.
Where upAPI wins
- The price is knowable in advance. A compute unit is consumed by a running container, so the cost of a job depends on how long it happened to run; a weighted unit is attached to the operation, so the same call always costs the same.
- Nothing you prepaid can expire. Apify’s own pricing FAQ says unused usage credits do not roll over and expire at the end of the billing cycle, so an unspent balance is money you already handed over. On upAPI, a call that returns nothing costs its declared weight and nothing else, and a monthly quota simply resets — there is no balance to lose.
- One contract and one owner. Every operation here is maintained by us, with one error envelope and one support address, instead of a marketplace of independently maintained actors of varying freshness.
- It runs as an MCP tool with no wrapper. Point Claude, Cursor or VS Code at the hosted server and the catalog is tools, authorized over OAuth rather than an API key in a config file.
What switching actually costs
Nothing to port. Operations are POST requests to `api.upapi.io/{slug}` with an `X-Api-Key` header and a JSON body, so a call is one HTTP request from wherever your actor client already runs. The free tier is enough to compare outputs on your own inputs before you decide.
Other comparisons: Bright Data, ScraperAPI, Serper, ApyHub, RapidAPI, ScrapingBee, Firecrawl, ScreenshotOne.