OpenAI Releases Three New Realtime Audio Models for Live Voice Applications

OpenAI released three new realtime audio models on May 8, 2026: GPT-Realtime-2, GPT-Realtime-Translate, and GPT-Realtime-Whisper, available through its Realtime API. The models enable developers to build reasoning agents with live voice, speech translation across 70+ languages, and streaming transcription capabilities for live voice applications.

Three new audio models

• ▪The three OpenAI realtime audio models are named GPT-Realtime-2, GPT-Realtime-Translate, and GPT-Realtime-Whisper.
• ▪The three OpenAI audio models are purpose-built for live voice applications.
• ▪OpenAI released three new realtime audio models on May 8, 2026.

Realtime API capabilities

• ▪OpenAI's three realtime audio models are available in the Realtime API.
• ▪GPT-Realtime-Translate supports speech translation across more than 70 languages.

Developer applications

• ▪OpenAI's three realtime audio models enable developers to build reasoning agents.

1 source