Tagged rate limits · show all articles
Grok's API looks like any other OpenAI-compatible endpoint until you read what xAI actually documents — a distinct pricing cliff, a real-time X search tool no competitor's docs describe, and a versioning policy that silently reroutes old model names.
A long agent loop resends its own history on every step, pays a fixed tool tax each time, and risks a retry that resends it twice. That arithmetic, not model quality, sets the real ceiling on how far a trajectory can run.
A rate card prices an uncached token, a cache write, a cache read, and an output token charged whether or not the answer arrives whole. It says nothing about tool overhead, a retried request, or the safety check every response already passes through.