AI cost reporting

AI costs, by customer.

Compare customer revenue with AI usage costs. Review workflow spend and compare estimates with provider invoices.

Start free See how it’s calculated

Start free. No card required. Team is $400 per month.

OverviewExample data

$8,830.00priced spend

32.1%contribution margin

attributedunattributed

Day 1Day 14

Reconciled against anthropic. $8,830.00 calculated against $8,830.00 billed. Ties out exactly.

The product

Customer revenue and AI cost, together

Track cost by customer, feature and workflow. Import customer revenue to review AI contribution margins. The view below uses example data.

CustomersExample data

$13,000.00revenue

$8,830.00AI cost

$310.00unattributed

62.0%from one workflow

Cost by customer

Customer A$5,200.00

Customer B$1,940.00

Customer C$962.00

Customer D$418.00

Unattributed$310.00

Margin by customer

Customer	Pays	Costs	Margin
Customer A	$8,000.00	$5,200.00	35.0%
Customer B	$3,500.00	$1,940.00	44.6%
Customer C	$1,200.00	$962.00	19.8%
Customer D	$300.00	$418.00	-39.3%

Customer D has $300.00 in revenue and $418.00 in AI cost, a $118.00 shortfall in this example.

Illustrative data. For calculation details, the methodology page states how each one would be calculated from usage and revenue.

The ingest call

Connect usage through a single endpoint

Send normalized JSON from your application or gateway. Cupelio receives usage outside the request path. Include a stable external_id to deduplicate retries.

You send

curl -X POST https://cupelio.com/api/ingest \
  -H "Authorization: Bearer cup_live_..." \
  -H "Content-Type: application/json" \
  -d '{
    "source": "generic",
    "events": [{
      "external_id": "req_01H8XYZ",
      "payload": {
        "provider": "anthropic",
        "model": "claude-sonnet-5",
        "input_tokens": 1000000,
        "output_tokens": 1000000,
        "customer": "customer-a",
        "feature": "document-analysis"
      }
    }]
  }'

Cupelio stores

model	claude-sonnet-5
input	1,000,000 at $2.00 / 1M
output	1,000,000 at $10.00 / 1M
customer	customer-a
cost	$12.00

{
  "accepted": 1,
  "duplicates": 0,
  "priced": 1,
  "unpriced": 0,
  "unattributed": 0,
  "not_usage": 0,
  "failed": 0
}

Example calculation: $2.00 input cost plus $10.00 output cost equals $12.00. Prices are selected by the event date. This request illustrates the normalized JSON contract; it is not a live request. See the methodology for current price records.

Four steps

From usage to customer margins

Connect usage, include customer identifiers and import revenue. Use the same identifiers throughout to match cost with revenue.

01
Connect usage
Create an ingest key and send normalized usage events from your application or gateway. Follow the setup guide directly or with your AI assistant.
02
Identify customers
Include your customer identifier on each event. Add feature and workflow identifiers to compare how AI is used. Events without a customer stay in a separate unattributed category.
03
Import revenue
Upload revenue as CSV using the same customer identifiers. Compare revenue with priced AI usage to calculate each customer's AI contribution margin.
04
Compare invoices
Record the provider invoice total for the period. Review the signed difference between billed charges and estimated usage cost.

The difference

Review cost with business context

Compare estimates with invoices

Record the provider invoice for the reporting period. Cupelio compares it with estimated usage cost and shows whether the totals match.

Compare model costs

Review what the same token volumes would cost on alternative models. These comparisons estimate cost only; model quality and performance require your evaluation.

Reporting you can inspect

Clear inputs. Traceable calculations.

Understand how usage is received, how costs are calculated and how invoice differences are reported.

Usage ingestion

Submit normalized usage to Cupelio with an ingest key. Your provider credentials are not required.

View security

Published methodology

Review the cost formula, effective-dated model prices and reporting limits behind every estimate.

View methodology

Visible invoice differences

Compare estimated usage cost with the provider invoice. Differences remain visible and are not silently distributed across customers.

View calculations

Flat monthly pricing

Team is $400 per month, with no percentage of your AI spend.

View pricing

Start with your usage data

Create a workspace, connect usage and explore customer cost reporting. Start with sample data or follow the ingestion guide.

Start free View methodology