OpenAI Launches GPT-Realtime-2 Voice Model, Real-Time Translation and Transcription Tools in Updated API

OpenAI announced new voice intelligence features for its API on Thursday, launching three models designed to let developers build applications that move beyond simple call-and-response audio interactions. The company launched GPT-Realtime-2, a voice model built to create realistic vocal simulation that can converse with users.

GPT-Realtime-2 is built with GPT-5-class reasoning and was created to deal with more complicated requests from users.

5. OpenAI also launched GPT-Realtime-Translate for real-time translation services that keep pace with the user conversationally. GPT-Realtime-Translate supports more than 70 input languages and 13 output languages.

The company launched GPT-Realtime-Whisper, which provides live speech-to-text transcription capabilities captured as interactions occur. All of the new voice models are included in OpenAI’s Realtime API. “Together, the models we are launching move real-time audio from simple call-and-response toward voice interfaces that can actually do work: listen, reason, translate, transcribe, and take action as a conversation unfolds,” OpenAI said.

Translate and Whisper are billed by the minute while GPT-Realtime-2 is billed by token consumption. TechCrunch reported that the new features could be handy for customer service systems. OpenAI notes that its new features will assist with education, media, events, and creator platforms, among others.

The company said it has built guardrails to stop its new features from being abused to create spam, fraud, or other forms of online abuse. OpenAI embedded certain triggers in the system so that conversations can be halted if they are detected as violating our harmful content guidelines. TechCrunch Disrupt 2026 is scheduled for October 13-15, 2026 in San Francisco, CA.