Voice and music · by OpenAI
OpenAI Realtime
Talk with a voice assistant that takes actions mid-conversation.
speech, text, image → speech, text, data · speech → speech, text · speech, text, image → speech, text, events
- Way in
- Official APISign up yourself, api key
- To start
- $$
- Runs
- In the cloudAlso through Azure, OpenRouter
- Checked
- Nov 3, 2026Prices as of that day
What it can do 7
Hold a low-latency spoken conversation with tool calls (speech in, speech out)session max duration not stated in guidespeech, text, image → speech, text, dataaudio $32/M in, $64/M out; text $4/$24 (gpt-realtime-2.1); mini $10/$20 audio
Live speech translation (gpt-realtime-translate)speech → speech, text$0.034/min
Hosted GPT-Live voice sessionsspeech → speech, text$0.05/min + backend model and tools
Hold a realtime spoken conversation with tool calls (gpt-realtime-2.1)speech, text, image → speech, text, eventsaudio $32/1M in, $64/1M out (2.1); mini $10/$20
Generate speech-only output from text via the realtime/audio models (gpt-audio-1.5 / mini)text → speechaudio out $64/1M (gpt-audio-1.5), $20/1M (mini)
Translate live speech into another language (gpt-realtime-translate)speech → speech, text~$0.034/min
Transcribe live audio (gpt-live-transcribe / gpt-realtime-whisper)speech → text~$0.017/min
Does the same job
Prices and limits as of Nov 3, 2026, from OpenAI’s own pages. They change; check before you rely on them.Not affiliated with OpenAI. Not advice.Something wrong?