Requesty
Every Requesty request, priced and attributed.
Cloud Intelligence™ imports your Requesty spend and token usage every six hours, by model, hosting provider, API key and routing policy. Right next to the rest of your AI and cloud spend.
Read-only management key · 12 months of history · Refreshed every 6 hours
Requesty · October
what we import
Requesty data, in the dimensions you already use.
Every charge maps to standard Cloud Analytics dimensions, plus Requesty labels for model, hosting provider, routing policy, API key, BYOK and organization.
Requestymy-org9f2c…Model Inferenceopenai/gpt-4.1-nanowhat you can do with it
From the Requesty bill to a decision
Cost by API key
Slice by API key, model, hosting provider or routing policy.
Policy-routed spend
See how much spend fallback and load-balancing policies route.
Cost vs list cost
What you pay with the credit markup, next to list price.
GenAI Intelligence
Requesty next to your other AI providers in one dashboard.
Budgets and alerts
Guardrails on gateway spend before credits run out.
Allocations
Put Requesty spend on teams and products, by API key.
live in hours
Connect Requesty in three steps
Create a management key.
Manage set to Read, Completions to None, with a tiny spend limit.
Connect in the catalog.
Name your organization, paste the key, then test the connection.
Reports in hours.
Up to 12 months of history imported. We email you when it's ready.
We only read usage, never send requests. Disconnect any time.
Start freeFrequently asked
questions
What key does the connection need?
A Requesty API key with Manage permission set to Read and Completions set to None. Requesty's usage API rejects keys without Manage permission, and we suggest a monthly spend limit of 0.01 as a safeguard.
What does the imported cost represent?
Cost is the credits drawn plus Requesty's 5% markup on credit purchases, so it matches what you pay. List cost is the credits drawn at provider list price, as shown in Requesty's analytics.
How is bring-your-own-key usage handled?
BYOK requests are billed by the upstream provider, so they carry zero cost in Requesty data to avoid double counting. Their estimated provider spend is available as the genai.cost.byok_external metric.
How much history is imported, and how fresh is it?
We import up to 12 months of history on connect and refresh every 6 hours. Requesty reports daily in UTC, and a day is imported once it has closed.
Which dimensions and labels can I use?
Billing Account is your organization name, Project is the API key ID and SKU is the hosting provider and model. Labels add model, provider, routing policy, API key, BYOK and organization.
Are there any data caveats?
Failed requests appear in request counts with zero cost. Prompt-cache tokens are included in input tokens because Requesty doesn't report them separately.
Connect Requesty. Know what every route costs.
Free to start. Read-only. Reports in hours.