GPT 5 mini for streaming avatars?

I want to switch to GPT 5 mini for many reasons, cost, speed, more context, etc.

will the streaming avatars support this? I switched to web speech api from a custom VAD setup for STT and it cut my response time by about 3 seconds. I’m now down to about 4-5 second response times with in depth responses. My hope was to get another 1-2 seconds cut by switching to gpt 5 because the context responses are taking 3 full seconds.