> ## Documentation Index
> Fetch the complete documentation index at: https://narev.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.
# Introduction
> Narev maintains real-time LLM pricing. Pull rates over REST or MCP, bill AI SDK usage, or ship the catalog into your warehouse.
> For AI-driven implementation, read the following resources immediately:
>
> 1. **Model Context Protocol (MCP) server:** `https://www.narev.ai/docs/mcp` (search and fetch documentation).
> 2. **Skills:** `npx skills install narevai/skills` (execute specific workflows).
> 3. **Index of all pages:** `https://www.narev.ai/docs/llms.txt` (fetch documentation and guides, identify relevant Cloud, SDK, or Self-Hosted paths, then read full pages before answering).
Narev maintains **the most carefully maintained**, real-time LLM pricing. You can use it to:
* deploy your own AI gateway
* bill your users per token, request, or usage
* run AI cost observability platforms
## Problems Narev solves
LLM prices change hourly now. Unless you scrape them in real time, you're behind.
Narev keeps a live price catalog you can trust and bill against.
Token counts alone don't tell you cost per request.
Either send us a request with token counts to `/calculate` or pull our data to your pricing map.
Lago, OpenMeter, Stripe, Polar don't maintain LLM pricing.
Narev has the most carefully maintained real-time LLM pricing and a set of tools to help you send dollar usage.
You need a price map you own, not whatever estimate another gateway baked in.
Set custom prices in Narev Cloud and push those rates into your gateway and tooling.
Token counts alone don't tell you cost per request.
ai-billing middleware resolves cost on the generation and can show usage in the UI.
## Getting started
### Bare metal
`GET` providers, models, and prices. `POST` cost calculation. No SDK required.
Wire an agent to `https://mcp.narev.ai` for live prices and docs.
If LLM prices live as constants in a repo, the app opens PRs when rates change.
### TypeScript
Wrap `generateText` / `streamText` with `@ai-billing/*` middleware and send usage to Polar or Stripe.
Provider middleware, price resolvers, and billing destinations in `@ai-billing/core`.
## Building something bigger?
When the public API isn't enough, Narev ships the catalog into your pipelines.
Stream price updates as rates change. Keep gateways and caches hot without polling.
Land full catalog dumps in object storage for your own loaders and lakes.
## Working with a data warehouse?
Managed ELT. Sync the Narev price catalog into Snowflake, BigQuery, or Redshift on a schedule.
Open-source or Cloud ELT. Pull catalog rows into the warehouse you already run.
Singer-based pipelines. Version the extract and load of Narev prices in git with your other jobs.
This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.