Cost Optimization

How can I reduce text API cost?

Use smaller models for draft generation, shorten prompts, trim long conversation history and cache reusable answers when possible.

How can I reduce image or video cost?

Validate prompts on cheaper models or smaller settings first, then run the final high-quality generation once the workflow is stable.

How do I monitor cost effectively?

Split keys by project, review key-level usage regularly and compare model usage patterns before changing routing or defaults.

Practical Checklist

1. Use a cheaper fast model as the default
2. Switch to stronger models only for hard tasks
3. Set max_tokens conservatively
4. Cache repeated outputs
5. Review project-level usage weekly