Deep diveProvider guide
Latency optimization
Generate fewer tokens: ask for only the detail and structured fields the app uses, and lower verbosity or reasoning effort for routine work. Combine sequential calls when one request has what it needs, and run independent calls in parallel.
Dossiers
- OpenAIthis guide's dossier