The number of tokens retrieved from cache.
The number of completion tokens generated.
The number of invocations.
The name of the model.
The number of prompt tokens used.
The total number of tokens used.
The number of tokens retrieved from cache.