Pricing API quickstart

You need curl and network access to https://api.narev.ai. These requests are free and require no authentication. They do not execute inference.

Find a model

Search the pricing catalog for a model:

curl --fail-with-body -G 'https://api.narev.ai/v1/prices/search' \
  --data-urlencode 'q=gpt-4o'

Read the returned model_id and provider_id. The same model can have different prices across providers.

Retrieve provider prices

Use the identifiers from your search to select a price row:

curl --fail-with-body -G 'https://api.narev.ai/v1/prices' \
  --data-urlencode 'provider_id=openai' \
  --data-urlencode 'model_id=gpt-4o'

Token rates in pricing are USD per token. Multiply a token rate by 1,000,000 to display a price per million tokens. Rates can change; use the returned values rather than copying a fixed example total.

Calculate request cost

Supply estimated or recorded token counts:

curl --fail-with-body 'https://api.narev.ai/v1/traces/cost' \
  -H 'Content-Type: application/json' \
  -d '{
    "model_id": "gpt-4o",
    "provider_id": "openai",
    "usage": {
      "prompt_tokens": 1000,
      "completion_tokens": 500
    }
  }'

Read cost_breakdown.total for the calculated USD cost. For workloads with caching, reasoning, or other billable dimensions, supply the applicable usage fields from the cost reference. A failed lookup is not a zero-cost request.

Next steps

Connect an MCP client to query pricing from your agent.
Read the API reference for filters, pagination, and response fields.
Track usage with AI-billing inside your application.

Get started

API Reference

Model pricing

Pricing API quickstart

Find a model

Retrieve provider prices

Calculate request cost

Next steps

​Find a model

​Retrieve provider prices

​Calculate request cost

​Next steps

Find a model

Retrieve provider prices

Calculate request cost

Next steps