The reason the model stopped generating. Common values include "stop" (natural end), "length" (token limit reached), and "tool_calls" (the model invoked a tool).
The final assistant message produced by the completion.
The complete message history for the conversation, including all user, assistant, and tool messages in order.
Optionaltoken_Token consumption breakdown keyed by model ID. Each value is a map with "input_tokens" and "output_tokens" counts. null when usage data is unavailable.
The result of an AI chat completion request. Returned by chat completion endpoints after the model finishes generating.