Calls used to run through three separate services for speech recognition, reasoning, and voice generation. They've been replaced with a single model, gpt-realtime-2.
Nothing to do on your end — call and text as usual.
Calls used to run through three separate services for speech recognition, reasoning, and voice generation. They've been replaced with a single model, gpt-realtime-2.
Nothing to do on your end — call and text as usual.