May 8th news, todayOpenAI Release three in real timeVoice ModelThree scenarios, one for voice reasoning, one for real-time translation and one for flow:
GPT-Realtime-2: Build voice intelligence for the environment. They allow for more in-depth reflection, operationalization, treatment of interruptions and the continuation of the dialogue
GPT-Realtime-Translate: supports real-time translation in more than 70 input and 13 output languages, breaking language barriers and helping people communicate more naturally
GPT-Realtime-Whisper: Real-time audio stream, production of subtitles and notes。
Among them, GPT-Realtime-2 carries the "GPT-5 level reasoning capability" designed specifically for voice-interactive scenarios, capable of processing complex requests during ongoing dialogue, multi-wire calls for external tools, responding to user interruptions and maintaining the natural flow of dialogue。
All three models are open to developers through OpenAI Realtime API and can be tested in OpenAI Playground。