Models & Receipts
Models & Receipts
GET /v1/models and GET /v1/receipts
GET /v1/models
Requires an API key. Returns every model in the catalog with its per-endpoint capabilities: context window, max and default output tokens, and current pricing as 12-decimal-place per-million-token strings.
A model entry looks like this:
Each model’s capabilities array has one entry per endpoint it serves — a model with both chat completions and responses capability has two entries, each with its own limits and pricing.
This endpoint takes no query parameters — any query string returns a 400. Responses are sent with cache-control: no-store, so what you get back is always current: pricing, limits, and the model list itself.
GET /v1/receipts/{receipt_id}
Requires an API key, and only resolves receipts belonging to your own account — an unknown or foreign receipt ID returns 404 receipt_not_found inside the standard error envelope.
A receipt looks like this:
state progresses through the request’s lifecycle to a terminal state. usage and total_cost_usd are null until the receipt reaches that terminal state — poll or wait before reading the final numbers.
For a full field-by-field glossary of the receipt object, see Receipts & Billing.