cheapinference.dev
A LITTLE LESS SPEND

Simple math.
Better economics.

Prepaid inference at 20% below OpenAI list price. No subscription. No monthly commitment. Just credit, ready when you are.

PAY AS YOU GO
$80Cash paid
$100Usage value

For every $1 you pay, you receive $1.25 of usage credit. Requests deduct credit at the model’s list rate.

Get started

Registration is open. Purchase availability depends on your account.

Everything you need.
Included from day one.

  • Supported OpenAI text and embedding models
  • Streaming and client-executed function tools
  • Shared organization credit and member roles
  • API keys with monthly usage limits
  • Token, model, key, and request-level tracking
  • Refund requests for unused credit
NO GUESSWORK

What happens when you make a request?

01

Credit is reserved

We reserve enough available credit for the request before it reaches the provider.

02

Usage is measured

Provider-reported input, cached, cache-write, and output tokens determine the final usage value.

03

The balance is settled

Actual usage is deducted once and unused reservation is released. Your usage page records the result.

If a successful upstream request has no usage report, the reserved estimate is charged and marked for review. See billing and usage for details, rounding, and credit validity.

A model for every workload.

Compare effective input and output rates in the model catalog.

Explore models