Measure cost per useful result
- Input tokens
- Output tokens
- Retries
- Tool calls
- Human correction time
- Failed or abandoned generations
Use budget controls early
Set limits at project and API-key level before traffic grows. A low token price does not protect an unrestricted key from loops or accidental concurrency.
Use the live pricing page
Model rates can change. The pricing page reads the active catalog and should be treated as the source of truth for current public rates.
Common questions
Where do model availability and prices come from?
Public model pages read the active catalog exposed by the service. A model is not presented as available merely because its name appears in an article.
Can I limit spending by project or API key?
The existing product supports project and key controls, including budget and rate constraints. Exact fields and enforcement behavior are documented in the console and API documentation.