observability-openai-cost-calculat-when-using-openai-o3-mini-or-o1-models-the-cost-ca-1d4e6eea

THEN Modify the cost calculation logic in the callback handler to retrieve reasoning tokens from usage_metadata['output_token_details']['reasoning'] and add them to the completion_tokens before computing the completion cost. This ensures billing matches OpenAI's pricing structure for reasoning models.

if 'reasoning' in usage_metadata.get('output_token_details', {}):
    completion_tokens += usage_metadata['output_token_details']['reasoning']