hi guys i'm trying to connect Gemini with VAPI creating my own server that run the transformation process. i already create a code that get the openai call from vapi, transform that call into gemini, then gives back the response in openai format, the thing is. how can i use it then? i have the url to use, but im stuck becaus if i select custom LLM, and the i put that url in the url text box, it does anything. i mean it says the initial message but when i try to talk the ai doesn't say anything, not even stop the call, just silence. anyone else is having this problem too?