Inference supply reference

Use this when choosing models for a model-routing quote or checking which enabled providers can carry the work. OpenAI and Anthropic entries default to their discounted async batch inference lanes when those providers publish batch rates.