[BUG] max_tokens parameter error when evaluating with OpenAI models

[JoonYong-Park]

Describe the bug
Evaluation runs fine with gpt-4o-mini, but fails with a 400 Bad Request when using the new gpt-5-nano-2025-08-07 model.

Root Cause
According to the official documentation, the max_tokens parameter is now deprecated and incompatible with newer models (like the o-series and gpt-5). The API now requires max_completion_tokens instead.

Error Log

httpx.HTTPStatusError: Client error '400 Bad Request' for url '[https://api.openai.com/v1/chat/completions](https://api.openai.com/v1/chat/completions)'
For more information check: [https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/400](https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/400)
ERROR - openai_backend - OpenAI attempt 3/3 failed: Error code: 400 - {'error': {'message': "Unsupported parameter: 'max_tokens' is not supported with this model. Use 'max_completion_tokens' instead.", 'type': 'invalid_request_error', 'param': 'max_tokens', 'code': 'unsupported_parameter'}}

Official Reference

• Docs link: OpenAI API Reference - Chat Completions
• Screenshot: