Usage analytics
See what your AI traffic costs,day by day.
The console overview adds up requests, tokens and estimated cost for each project, with a daily chart. Pull the same numbers into your own dashboards and billing with the Usage API.
How analytics work
Every request is counted as it happens. You choose the period and the traffic to look at.
- 01
Every request counted
Each request is logged with its model, tokens and estimated cost, priced from the model’s per-token rates.
- 02
Pick a period
View today, yesterday, the last 7 or 30 days, or 3 months. Paid plans can also view a billing cycle.
- 03
Split the traffic
Show all traffic, production only, or Playground runs, so testing doesn’t blur your production numbers.
- 04
Take the numbers with you
The Usage API returns a project’s usage for a month or a quota cycle, with a breakdown by day.
The numbers
- Requests · tokens · costWhat the overview adds up for each project
- DailyChart of requests and tokens for the period you pick
- 24 monthsHow far back the Usage API goes
- 60 secondsMost a Usage API result can lag behind live traffic
The Usage API
Call it with a project API key. Ask for a calendar month, or for the project’s current or previous quota cycle.
GET /v1/usage?month=2026-10
Authorization: Bearer mr_live_…{
"month": "2026-10",
"billable_requests": 48210,
"requests": 51377,
"failed_requests": 312,
"prompt_tokens": 18440219,
"completion_tokens": 6120884,
"price": 41.87,
"by_day": [
{ "date": "2026-10-01", "billable_requests": 1490, "requests": 1602, … },
…
]
}Billable and total, side by side
Billable requests leave out test mode, Playground and session messages, so you see exactly what your plan counts.
Bill your own customers
Give each customer their own project, read its usage each month, and charge for it. On paid plans you can also cap each project with a monthly quota.
Sessions included
Usage also reports active sessions, messages appended to sessions, and signals sent.
Good to know
Cost is an estimate
Prices come from each model’s per-token rates, not from your provider’s invoice. Your provider bills you directly.
Billing cycles on paid plans
Rolling periods work on every plan. Viewing usage by billing cycle needs a paid plan.
One project per key
The Usage API only returns the project that owns the API key, so a key can’t read another project’s usage.
Cache savings on each workflow
Each workflow’s cache dashboard shows its cache hits and the money they saved.
Frequently asked questions
What does the overview show?
Total requests, prompt and completion tokens, estimated cost and active API keys for the project, with a daily chart of requests and tokens.
How is cost calculated?
From each model’s published per-token prices and the tokens used. It is an estimate; your provider bills you directly for its usage.
Can I see usage for a billing cycle?
Yes, on Pro and Scale. Free plans can use rolling periods such as the last 7 or 30 days.
How do I get usage into my own tools?
Call GET /v1/usage with a project API key. Pass a month like 2026-10, or ask for the current or previous quota cycle.
Do test requests show up in usage?
They are logged, but they are not billable. The Usage API reports billable requests and total requests separately.
Works well with
See your first day of usage
Start on the free plan and watch requests, tokens and cost add up in the overview.