Description of the feature request:

I am using OpenAI Realtime API, and I could directly connect the client with OpenAI server using webRTC and ephemeral token. Whereas, in gemini live API, I have to setup the websocket proxy server and websocket browser client to do the same.

I would really like to have the gemini live API support direct webRTC connection similar to OpenAI. This way the latency and scalability issues are very well addressed. also, I don't have to worry about signaling server needs if it is directly supported by gemini.

What problem are you trying to solve with this feature?

latency, reliable and scalable way of building apps using gemini live API, through webRTC implementations.

Any other information you'd like to share?

No response