OpenAI released GPT-Live-1.1 on 15 October 2026, an update to its voice features that adds new voices in Japanese, Korean, Hindi, Mexican Spanish, German, and French. In a post on X, the company says the update also improves accents and reduces what it calls “checking,” a term it does not define.
The update covers Voice in ChatGPT, Codex, and dots, and OpenAI says it is rolling out to all users from that date. Developers get less. API access is listed as “coming soon,” with no date attached, so nobody can yet build a Hindi or Korean voice product on these voices.
This is a voice refresh, not a new model generation, and the post is an announcement rather than a result. It includes no audio sample, no latency figure, no measure of how far “checking” fell, and no named evaluation. “Better accents” is OpenAI’s own description.
For teams building voice products, the open question is the API date. Until OpenAI gives one, the new languages are a consumer feature.
Based on OpenAI’s post on X (OpenAI on X) of 15 October 2026.
