Our realtime/sessions request was successful, but gpt-4o-mini-realtime and gpt-4o-realtime aren’t responding with audio.
This was working 4 hours ago, and we’re not hitting API rate limits as well.
Our request body:
{
model: "gpt-4o-mini-realtime-preview-2024-12-17",
voice: "sage", // Use the voice from query params
instructions: "some-long-prompt-here",
input_audio_transcription: {
model: "gpt-4o-transcribe"
},
turn_detection: {
silence_duration_ms: 1000,
type: "server_vad"
},
}