The OutputTokens property returns the number of output tokens
generated by the most recent response received from the provider.
Output tokens represent the content generated by the model, including
text, reasoning content and other generated data returned as part of
the response.
Output token usage is typically reported by the provider as part of
the response metadata. If token usage information is not available,
the property value may be zero.
Generated output tokens are included when calculating usage limits
and billing costs for most providers. Applications which generate
long responses or use reasoning-oriented models may consume
significantly more output tokens than expected.
To determine the estimated number of tokens currently used by the
conversation history, check the value of the ContextUsed
property.