It's the same pipeline just using voice to send the prompt and I checked Whisper output and it's the same thing as what I was typing, but through voice it gives me bad answers more than half the time...
#Same assistant using text input nails every prompt but same prompt using Voice Preview mostly fails
merry coral