Skip to content
SFAI
SFAI / PRACTICAL GUIDE

Prompt caching and batch pricing: comparison checklist

Discounts depend on the offer’s conditions and your workload.

Identify the pricing components

A provider may quote input, output, cache-read and cache-write rates separately. Not all offers report every component. Missing cache pricing is unknown, not a zero-cost feature. The exchange keeps the published values attached to the provider offer.

Separate batch from standard service

Batch processing can have different timing and pricing conditions. Do not compare a batch rate against an immediate-response rate without acknowledging the service difference. Model variants and service tiers remain separate so a cheap rate is not mistaken for the same product.

Estimate achievable savings

Use your expected reusable prompt volume and the provider’s cache eligibility and retention rules. A discount is only useful if the workload qualifies. Confirm the current documentation and billing behavior with the provider; the exchange calculator is a planning aid, not a billing guarantee.

Apply the method to current data

Browse current source-qualified model profiles · Review sources and publication limits