Reference · Glossary
Token budget
A **cap on tokens** per request, user, or day — prevents runaway spend from long contexts or agent loops.
When to use
Production agents, free tiers, and batch jobs with fixed margins.
When not to
Research notebooks where cost is irrelevant — still useful as a sanity check.
Example
Hard stop at 8k output tokens per chat; daily org budget alert at 1M tokens.