Skip to content
context New models

OpenAI releases GPT-Live-1.1 model for multilingual voice conversations

clearly official source

OpenAI has released the GPT-Live-1.1 model with improved pronunciation, accent, and transcription for multilingual voice conversations, and added standby mode to pause inference during inactivity.

OpenAI has released the GPT-Live-1.1 model, the successor to the GPT-Live-1 model intended for voice conversations in the v1/live/sessions API. According to the company, the new version improves pronunciation, accent consistency, and transcription in multilingual voice conversations. The model is compatible with existing Live session configurations, so switching to the new version does not require major changes to the integration.

Along with the new model, OpenAI introduced standby mode for WebRTC sessions, available for both GPT-Live-1 and GPT-Live-1.1. Standby allows inference to be paused during periods of inactivity in a conversation and automatically resumed as soon as the user starts speaking.

The source materials do not state pricing, availability for specific regions, or a quality comparison with the previous version beyond the improvements described.

What changed

Why it matters

For developers of voice applications, this is a small but directly usable update: better pronunciation and transcription in foreign languages increases usability for international deployments, while standby mode reduces inference costs during periods when the user is not speaking, without requiring any changes to the application architecture.

Release card

GPT-Live-1.1

OpenAI

Inputs
voice conversations (audio) with transcription
Availability
OpenAI API, v1/live/sessions endpoint, compatible with existing Live session configurations
According to the sources, it is suitable for
  • multilingual voice conversations with emphasis on pronunciation and accent
  • transcription of voice conversations in Live sessions API
  • WebRTC session with standby mode that pauses inference during inactivity and resumes it when speech begins

According to the source, the model improves pronunciation, accent consistency, and transcription in multilingual voice conversations compared to the previous GPT-Live-1 version.

The card summarizes information from the article and any dated corrections, with a link to the original source. It is not our assessment of the model. It does not yet have a dedicated editorial profile. Model selection and other announcements →

Two audiences, two different impacts

What this means

01

For individuals

Developers working with OpenAI's voice API can update the configuration of an existing Live session to the new model without major changes and gain better pronunciation, accent consistency, and transcription in multilingual conversations.

What to do Developers using the Live API can switch to gpt-live-1.1 in the session configuration and enable standby mode for the WebRTC session.
More practical updates →
02

For a business

Companies building products on OpenAI's voice API can use standby mode to reduce inference costs during idle periods of a conversation while also improving output quality in multilingual scenarios, which is relevant for customer support, assistants, and similar deployments.

Development
What to decide Companies running voice applications with multilingual support can evaluate switching to the GPT-Live-1.1 model and enable standby mode to reduce inference costs.
More business impacts →
GPT-Live-1.1 voice API Live API multilingual OpenAI voice conversations

Check the original

Event sources

clearly official source · 1 publisher, 0 independent. We count feeds from the same owner only once.

1
OpenAI API Changelog primary source · first detected [GPT-Live-1.1](https://developers.openai.com/api/docs/models/gpt-live-1.1) improves pronunciation, accent consistency, and transcription in multilingual voice conversations. Use…