Pricing and usage
Understand token rates, request-priced models, and usage records.
Token-priced models
The public catalog shows input and output prices in USD per one million tokens. Input tokens include the content you send; output tokens are generated by the model. Models can price these two directions differently.
Request-priced models
Some image, video, audio, or specialized endpoints are billed per request rather than per token. Their cards show a request price. Resolution, duration, or other parameters may affect model-specific billing when documented.
Published price and account charge
The model page reads the public gateway catalog. Your dashboard is the authoritative record for account balance and completed request deductions. Before a large batch job, make a small request and confirm the resulting usage record.
Keep costs predictable
- Set sensible output-token limits.
- Use smaller or faster models for routine classification and extraction.
- Test prompts on a small sample before batch processing.
- Use separate API keys for projects that need independent usage tracking.