Guides
How AI API relays work, and how to check one
Written from our own measurements against this catalogue and the providers’ own documentation. Where a number appears, the source and the date it was read appear with it.
Going deeper
Canary-testing an AI API relay: a method you can run yourself
You cannot read a relay’s routing table. You can send it a prompt whose answer only one model gives, and see what comes back.
11 min readModel authenticity: what happens when one model name has four routes
One model name can sit in front of several upstream channels. Which one answers is a policy decision, and the policy is the thing worth reading.
12 min readWe audited our own list prices against the providers, and two were wrong
The number a discount is measured against is supplied by the party claiming the discount. We read ours against the source and published what we found.
10 min readPer 1K or per 1M: the 1000x error waiting in every price comparison
Two prices, two units, one table. The mistake does not look like a mistake — it looks like a spectacular deal.
9 min readReading a provider’s list price correctly: the conditions nobody quotes
A single number on a pricing page is usually several numbers wearing a coat.
11 min readPrepaid credit or a monthly bill: which billing model fits your workload
The billing model is not an accounting detail. It decides what happens on the day something goes wrong.
9 min readWhat “OpenAI-compatible” actually has to mean
Compatible with what, exactly? The request body is the easy half.
12 min read
The numbers these guides use are public
Every model on this station, its per-token price and the provider’s own list price are on the pricing page, readable without an account. The guides explain how to read them.
8 guides · last updated 2026-08-30