Use case
GPT-6 Astra prompt caching
prompt caching notes for gpt-6-astra: the rate card, a worked volume, and when to send this ID. A sample of 1 million input tokens plus 200 thousand output tokens is about $4.00 on corouter (input $2.00, output $2.00).
Cache-read rate
Cache read for gpt-6-astra lists at $1.00 and is $0.200 on corouter per million tokens. Uncached input is $2.00.
Only the input that actually hits cache is billed at the cache-read rate. This page does not promise your prompt will hit.
Where this ID sits
GPT-6 Astra is the highest standard list rate in the GPT rows here. Send gpt-6-astra when the task is worth that output rate and you want the 1.05M window with image input.
Questions
- Which model ID should the request send?
- Send gpt-6-astra. Do not put a different name in the model field.
- What is the minimum top-up?
- Prepaid credits, $5 minimum top-up, no subscription tier.