OpenAI has rolled out GPT-Live-1.1, its latest voice model for the ChatGPT app — and with it, a broader, more regionally tailored set of voices across six languages, including Hindi, Japanese, Korean, French, German and Mexican Spanish. The update began distribution on October 15 and is rolling out gradually. No end date has been announced. Users in India and the US will see different voice options depending on account and region, a detail OpenAI explicitly notes in its release notes: “the voices you see may vary by region and account.” English users now get Bubbie, a new voice added alongside the existing nine (Arbor, Breeze, Cove, etc.), all of which previously leaned heavily on US or UK English phonetics. German speakers gain Sonna and Miko. French, Japanese, Korean and Mexican Spanish each receive at least one new voice. Accents are said to sound more natural across the board. Though OpenAI does not quantify how or where that improvement manifests. Searching behavior also shifts slightly: Live now uses a wider range of short verbal cues to signal active searching, replacing older, narrower prompts. The update follows late-September enhancements that let voice mode interact with calendars and reminders, but only after manual confirmation via tap. This time, the change is under the hood: a new model layer, not just new features. Two variants ship: the full GPT-Live-1.1 and GPT-Live-1.1 Quip, a lighter version optimized for speed and lower latency. Access depends entirely on subscription tier. Free users get limited, unnamed access to Quip only. ChatGPT Go ($20/month) includes three hours of Quip use per rolling 24-hour window. Plus ($20/month) unlocks three hours of the full GPT-Live-1.1. Pro ($100/month) raises that to 15 hours. The $200 Enterprise tier removes the cap entirely. Though OpenAI says abuse guardrails still apply. None of these quotas changed with this release. They’re unchanged from the FAQ published October 16. To try the new voices, users must update the ChatGPT app, then navigate to Settings > Voice. If nothing new appears, their account hasn’t been migrated yet. No public timeline exists for full rollout. The update lands as search traffic for ‘chatgpt app’ and ‘chatgpt free’ remains high, and as eight US states sue OpenAI over child safety claims, unrelated to voice functionality but adding regulatory pressure on how the company positions and gates access to its tools. The voice model itself is not available outside the official ChatGPT app. There’s no standalone ‘ChatGPT Voice’ download, no web-only voice interface, and no Android or iOS SDK for third-party integration. It runs only inside the app. And only for accounts that have received the backend switch.