InferenceIndex

Early access · Private preview

Concepts

Credits and billing

Calls spend credits. The dashboard shows the balance.

A Compiled Agent Model is much cheaper than a frontier model on the runs we have measured. Calls spend credits. The dashboard shows usage during early access. The API is a private preview.

A short credit meter beside a tall frontier meter

What the tests show

These are published test figures, quoted as tests. They are not a rate card.

  • The home page shows early repeat runs as 5–10× faster.
  • Repeat runs are often 10-100x cheaper on our tests. Savings depend on the workflow.
  • The repeat-cost essay measures one invoice reconciliation run at 1.5s and $0.00023 per run versus Opus 5.5 at 24.8s and $0.105 per run, roughly 450x.
  • The home page lists support ticket triage at about 0.5s and $0.00009 per repeat, versus about 10s and $0.012 for a frontier agent.

Where to look

Open https://inferenceindex-app.web.app and sign in. Credits for your own calls are shown in the dashboard.