OpenAI Provider Notes
Credentials
- Set
OPENAI_API_KEYin your environment or passapi_keytoOpenAI(...).
Embeddings
- Default online embedding model:
text-embedding-3-small. - For lower latency on large volumes, consider batching embedding calls in your own pipeline.
Model selection
- Synthesized answers use
modelconfigured on theOpenAIclient. Use smaller models for cheaper, faster results when appropriate.
» See also: docs/tuning/operation_mode.md