{
"$type": "site.standard.document",
"bskyPostRef": {
"cid": "bafyreiai2jmo4l5fadpk2hhpy7aehwtsdxqxla4dblnamf245ka4ovjvqe",
"uri": "at://did:plc:lk3jfj3zq4k4wxnk474axylu/app.bsky.feed.post/3mlc4rlihvmj2"
},
"path": "/t/new-realtime-voice-models-in-the-api/1380471#post_3",
"publishedAt": "2026-05-07T20:42:13.000Z",
"site": "https://community.openai.com",
"tags": [
"Realtime and audio",
"Using realtime models",
"Voice agents",
"Realtime translation",
"Realtime transcription",
"Realtime with tools",
"gpt-realtime-2 model page"
],
"textContent": "Additional Documentation and Guides related to this release:\n\n * Realtime and audio\nUpdated overview for choosing between voice agents, realtime translation, realtime transcription, and request-based audio APIs. It explicitly routes low-latency voice agents to `gpt-realtime-2`.\n\n * Using realtime models\nNew/updated prompting guide for `gpt-realtime-2`, including reasoning effort, preambles, tool policies, unclear audio handling, exact entity capture, and long-session behavior.\n\n * Voice agents\nUpdated guide for building speech-to-speech agents with `RealtimeAgent` / `RealtimeSession`, WebRTC, tools, handoffs, and guardrails.\n\n * Realtime translation\nDedicated guide for `gpt-realtime-translate`, including `/v1/realtime/translations`, WebRTC/WebSocket patterns, listen-along translation, conversational translation, and production checklist.\n\n * Realtime transcription\nDedicated/refreshed guide for `gpt-realtime-whisper`, streaming transcript deltas, latency/accuracy tuning, vocabulary guidance, and production checklist.\n\n * Realtime with tools\nGuide for function tools, remote MCP servers, and built-in connectors in Realtime sessions with `gpt-realtime-2`.\n\n * gpt-realtime-2 model page\n\n\n\n\nPricing info:\n\n * `GPT-Realtime-2`: `$32 / 1M` audio input tokens, `$0.40 / 1M` cached input tokens, `$64 / 1M` audio output tokens\n * `GPT-Realtime-Translate`: `$0.034 / minute`\n * `GPT-Realtime-Whisper`: `$0.017 / minute`\n\n",
"title": "New Realtime Voice Models in the API"
}