OpenAI API configuration error: "Max Output Tokens" parameter causes extraction failure #32
Description
Description
There is a configuration issue when using OpenAI models in the "Bring Your Own LLM" section that causes AI extraction to fail.
Current Behavior
When selecting "OpenAI" in the "Bring Your Own LLM" section, only recent GPT models appear to be available
These recent models seemingly don't accept the "Max Output Tokens" parameter as input
This causes an error during AI extraction
Expected Behavior
OpenAI models should either:
- Accept the "Max Output Tokens" parameter without error, or
- The parameter should be omitted/adapted for models that don't support it
Note
This needs further testing and confirmation
Steps to Reproduce:
- Go to "Bring Your Own LLM" configuration
- Select "OpenAI" as provider
- Select one of the available (recent) GPT models
- Attempt AI extraction with "Max Output Tokens" parameter set
- Observe the error
Possible Root Cause
Recent OpenAI models may use different parameter names (e.g., max_completion_tokens instead of max_tokens) or have removed support for output token limits in the API call format being used.