Paying for the same answer again
FAQ bots, retries and popular prompts generate identical calls all day.
Caching
Ask twice, pay once
Real traffic repeats itself. When the same prompt arrives again inside the cache window, we return the stored answer at once and bill nothing for it. No configuration needed.
Instant repeats
served straight from cache
Not billed
cached responses are free
On by default
nothing to configure
The problem
What is slowing you down today
Latency on easy questions
Users wait for a fresh generation of an answer you already have.
Cache layers to build and run
Rolling your own cache means keys, eviction and invalidation to maintain.
Capabilities
Everything you need, nothing you don't
Automatic matching
Prompts are matched after trimming, lowercasing and collapsing whitespace, together with the model and temperature.
Per-account isolation
Cache entries are scoped to your account. Nobody else's traffic can hit your cache.
Zero price on hits
Cached responses report cache_hit true and a price of zero.
Clear status
The response status field reads cached, so your code always knows.
Bounded freshness
Entries expire after the cache window, so answers do not linger indefinitely.
Visible in history
Cache hits appear in request history, so you can see how much traffic repeats.
Use cases
What teams run on it
01
FAQ and help bots
Common questions answered instantly.
02
Client retries
A repeated request returns the stored answer instead of a new bill.
03
Popular content
Trending prompts served at cache speed.
04
Development loops
Re-running the same test prompt costs nothing.
How it works
From sign-up to production
- 1
Send a request
The first call is generated and stored for your account.
- 2
Repeat it
The same prompt, model and temperature inside the window is a cache hit.
- 3
Check cache_hit
The response tells you it was cached and priced at zero.
Control
Built so you stay in charge
Keys you control
API keys are shown once at creation and stored only as a hash. Revoke any key instantly from Settings.
Spend you can see
Every request carries its own receipt: tokens in, tokens out, latency and price. Nothing is billed that you cannot trace.
Limits you set
Per-account rate limits and a free monthly allowance keep experiments safe before they become production.
Data you decide on
You choose exactly what you send. Repeat prompts can be served from your cache, and cache hits are never billed.
For developers
One endpoint. Every model.
Send a prompt and a capability, or name a model. One bearer key, one small JSON body, and a full receipt on every response.
curl https://YOUR-DOMAIN/api/ainf/inference \
-H "Authorization: Bearer $AI_INFERENCE_KEY" \
-H "Content-Type: application/json" \
-H "Idempotency-Key: order-1042" \
-d '{"prompt":"Summarise this ticket in one line","capability":"chat"}'Questions, answered
What counts as the same prompt?
The prompt after trimming, lowercasing and collapsing whitespace, plus the same model and temperature.
How long are responses cached?
For the account's cache window. After it expires the next call is generated fresh.
Can I avoid the cache for one request?
Change the prompt or the temperature and it becomes a different cache entry.
Do cached calls count against my allowance?
Cached calls are priced at zero. They still appear in history so you can see them.
Let repeat traffic pay for itself
Caching is on from your first request. Start free.