Fix image generation in OpenaiChat

[hlohaus]

No description provided.

[github-actions]

Pull Request Review

Author: H Lohaus

Title: Fix image generation in OpenaiChat

Thank you for contributing to the project! Your changes to the OpenaiChat provider look great and seem to address the image generation functionality effectively.

Summary of Changes:

• Added image_models to the class, which uses the existing fallback_models.
• Modified the create_async_generator method to set history_disabled to False by default.
• Removed unnecessary WebSocket handling code that was commented out.
• Improved the handling of image generation within the iter_messages_line method, ensuring that images are processed correctly when received.

Review Comments:

• The addition of image_models is a good enhancement, as it clarifies the purpose of the models being used.
• Setting history_disabled to False by default may improve user experience, but please ensure this aligns with the intended functionality.
• The removal of the commented-out WebSocket code helps clean up the codebase, making it easier to read and maintain.
• The changes in iter_messages_line enhance the handling of multimodal responses, which is crucial for image generation.

Overall, the modifications are well-structured and improve the functionality of the OpenaiChat provider.

Please let me know if you need any further feedback or assistance!

Best regards,
[Your Name]