OpenAI said Thursday that its API will include several new speech intelligence features designed to help developers create apps that can talk to users, transcribe and translate them.
The company’s new GPT‑Realtime‑2 is another voice model built to create realistic voice simulations that can converse with users. However, unlike the previous version (GPT-Realtime-1.5), this version is built on what OpenAI says is GPT-5 level inference and is intended to handle more complex requests from users.
The company will also launch GPT-Realtime-Translate, designed to provide a real-time translation service that “keeps pace” while literally talking to the user. This feature includes over 70 input languages (the languages you understand) and over 13 output languages (the language you communicate to the speaker).
Lastly, the company also launched GPT-Realtime-Whisper, a new transcription feature that provides users with real-time speech-to-text capabilities that are captured as interactions occur.
“Together, the models we are launching move real-time audio from a simple call and response to a voice interface you can actually work with – listening, inferring, translating, transcribing and taking action as the conversation unfolds,” the company said.
Who will benefit from this update? It’s an obvious goal for companies looking to expand their customer service capabilities. But OpenAI also points out that the new features will benefit a variety of areas, including education, media, events, and creator platforms.
Although these tools may seem useful from a business perspective, they can also be misused. The company said it has built guardrails to prevent the new feature from being abused to create spam, fraud or other forms of online abuse. “Certain triggers are built into the system so that conversations can be halted if they are detected to be violating our harmful content guidelines,” OpenAI said.
Tech Crunch Event
San Francisco, California
|
October 13-15, 2026
All new speech models are included in OpenAI’s Realtime API. Translate and Whisper are billed by the minute, while GPT-Realtime-2 is billed based on token consumption.
If you purchase through links in our articles, we may receive a small commission. This does not affect our editorial independence.
