OpenAI Launches Three Realtime API Voice Models Including GPT-Realtime-2
OpenAI shipped a simultaneous triple release in its Realtime API: GPT-Realtime-2 (the company's most intelligent voice model yet, with GPT-5-class reasoning capable of thinking and handling interruptions), GPT-Realtime-Translate (streaming translation across 70+ input and 13 output languages), and GPT-Realtime-Whisper (real-time audio transcription for live captions). Sam Altman called GPT-Realtime-2 "a pretty big step forward" and noted growing voice usage, particularly among younger users.
Why It Matters
The coordinated triple launch signals a deliberate industry push to make voice the dominant AI interface of H2 2026. Voice-native agent architectures will likely see accelerated adoption for customer service, accessibility, and real-time translation use cases.