@reustle

Thanks! 3 in parallel, 4o realtime audio model (basic listening) + whisper model (transcripts) + o3-mini (double-checking regularly)

0 reply

0 recast

2 reactions