// guide

How to Track OpenAI API Costs (3 Methods Compared)

OpenAI bills by the token, and token usage is invisible until the invoice arrives. A single misconfigured eval loop or a switch from mini to a frontier model can multiply your bill overnight. Here are the three ways teams actually track it.

1. The OpenAI usage dashboard

The built-in dashboard at platform.openai.com/usage shows daily spend, broken down by model and project. It's free and accurate — but it's manual. Nobody checks it daily, it doesn't alert you, and it can't see your Anthropic, Vercel, or infrastructure spend next to it.

2. The Usage & Costs API with an admin key

OpenAI exposes organization-level usage and cost endpoints that require an admin key (created at platform.openai.com → Settings → Organization → Admin keys — not a regular API key). With it you can pull daily cost buckets, token counts per model, and project attribution, then pipe them into your own spreadsheet or warehouse. This is the right raw material, but you have to build the dashboard, the history, and the alerting yourself.

3. Automated tracking (what Prism does)

Prism connects to that same admin API with a read-only key and does the rest automatically: daily syncs, per-model and per-project breakdowns, month-over-month trends, spend forecasting, and budget alerts before you blow through a limit. It also tracks the rest of your stack — Anthropic, Vercel, Railway, GitHub, and 100+ other dev tools detected from invoices — so AI spend sits in context, not in a silo.

Which should you use?

• Dashboard only — fine for hobby projects under ~$50/month.
• DIY admin API — right if you already have a data team and a warehouse, and AI spend justifies the build.
• Prism — right if you want per-model visibility, alerts, and your whole dev-tool spend in one place without building anything.