I would like to suggest a feature enhancement regarding support for custom models in OpenAI format requests. Currently, when utilizing OpenAI API integrations, the system automatically switches to alternative APIs like Claude or Gemini based on the model name, even if I wish to use a custom model with a unique identifier.
It would be highly beneficial if we could manually specify any custom model within the OpenAI format request without the system automatically switching to different APIs. This would be similar to how some other applications (like the one shown in the attached image) allow users to define and add any custom model while still adhering to OpenAI’s API structure.
In addition, it would be great if this functionality could keep user-defined models separate from automatic model name validation, allowing more flexibility when dealing with experimental or private models.
Thank you for considering this suggestion!
