Request counts are tracked by Zora. Token counts appear when the model response includes an OpenAI-compatible usage object. Live