How do requests and credits work?#

Every billed API call spends one credit from your plan’s quota, and most search and retrieval endpoints cost exactly that — one credit per call. When your plan quota runs out, a paid plan keeps working by charging the same call to your pay-as-you-go balance at $0.01 per request instead of failing. APITube also keeps your standard requests, SSE stream and WebSocket stream in three separate quota pools, so heavy streaming never eats your search allowance.

Credits are the same thing the API calls points: they are your plan’s request allowance for the period. Each billed call draws one point from the pool, the dashboard counts it, and when a pool reaches zero the API stops serving that kind of call until you have quota again or top up.

What is a credit and where is it stored?#

A credit (a point) is one unit of request allowance held on your account, not on an individual key. Your account carries four separate balances: a standard-request pool, an SSE-streaming pool, a WebSocket pool, and a pay-as-you-go cash balance. Every key you create shares the same account pools, so revoking or adding keys does not change how much quota you have — it only changes which key the spend is attributed to. Each billed call also bumps that key’s lifetime usage total, which is how the dashboard shows spend per key.

How much does one request cost?#

The default is one credit per call, and the price is per call, not per article. A page that returns a single article and a page that returns the full 250-article maximum both cost exactly one credit — raising per_page and paginating is the cheapest way to pull large result sets. A few endpoints differ: /v1/fact-check costs 5 credits because it runs a retrieval-and-reasoning pass, and the real-time streams bill per article delivered rather than per connection. The full per-endpoint breakdown lives in what counts as one request.

curl "https://api.apitube.io/v1/news/everything?title=electric%20vehicles&api_key=YOUR_API_KEY"

That single call deducts one credit from your standard pool — but only if it returns at least one article. A search that matches nothing comes back empty and is not charged, so you can tighten filters and retry a zero-result query for free.

Which pool does a request draw from?#

Each kind of traffic has its own pool, and each is gated on its own balance:

• Standard requests — search, retrieval, counts and directory lookups draw from your main request pool.
• SSE streaming (/v1/news/stream) draws from a dedicated streaming pool and bills one credit per article it pushes. See how streaming quota is counted.
• WebSocket streaming (/v1/news/ws) draws from its own WebSocket pool, again one credit per article delivered.
• Pay-as-you-go balance is real money, charged only when a plan pool is empty.

Because the pools are separate, running your SSE stream to zero does not block a normal /v1/news/everything search, and vice versa.

When does my balance get charged instead of my quota?#

Your plan quota is always spent first. On a paid plan, once the standard pool hits zero and you still have a pay-as-you-go balance, the next request is charged to that balance at one cent — $0.01 per request — so the call succeeds instead of returning an error. This fallback is why a request always counts as one charge; only the pool it comes from changes.

Common Questions#

• Do credits and points mean the same thing?
• Do test keys spend credits?
• Does a bigger page of results cost more?
• Why is my streaming pool separate from my search quota?

Do credits and points mean the same thing?#

Yes. “Credits” in the dashboard and “points” in the API refer to the same request allowance on your account. One billed call spends one point, except for the fact-check endpoint (5) and the per-article streams.

Do test keys spend credits?#

No. Calls made with a test key hit the live API but never draw down any pool, so you can preview request shapes and responses without touching your quota or balance. Only live keys are billed.

Does a bigger page of results cost more?#

No. On the search endpoints the price is one credit per call whether the page holds one article or the full 250-article maximum. Fewer, larger calls therefore cost fewer credits than many small ones.

Why is my streaming pool separate from my search quota?#

Streaming bills one credit per article delivered and can run continuously, so APITube keeps its SSE and WebSocket allowances in dedicated pools. That isolation means a long-running stream cannot silently drain the quota you need for ordinary searches.