Currently for "Generate Image" action for Retool AI resource, ChatGPT image generation is the only one supported.
I'm always looking for lower latency models (ultimately, the lower latency something is, the faster my team works by 100x). ChatGPT is not competitive with both time to first token and chat completion time compared to Gemini.
We're very image heavy, so being able to experiment with other models for Images like Google's Neo-Banana or MidJourney's official API without having to create custom AI APIs would be lovely.