OpenAI has started rolling out GPT-Live-1.1, an updated voice model designed to make conversations in ChatGPT sound more natural, particularly when users speak languages other than English. Announced on October 15, 2026, the update also introduces GPT-Live-1.1 Quip, a smaller version of the model, and expands the available voice options across French, German, Hindi, Japanese, Korean and Mexican Spanish, alongside a new English voice called Bubbie.
The update focuses on how ChatGPT speaks, handles different accents and keeps conversations flowing while performing tasks such as web searches. For users in India, the addition of two new Hindi voices, Neer and Leher, is especially relevant as OpenAI increases its focus on voice-based interactions and multilingual access. The rollout is gradual, so the new model and voices may not appear on every account immediately. Users can check the available options under Settings and then Voice.
Key Takeaways
- New voice model: OpenAI is rolling out GPT-Live-1.1 to ChatGPT Voice.
- Smaller model: GPT-Live-1.1 Quip is also being introduced as a lighter version.
- 13 new voices: The update adds voice options across six language varieties, plus Bubbie in English.
- Hindi support: Two new Hindi voices, Neer and Leher, are part of the rollout.
- More natural speech: OpenAI says the update improves accents and makes voice interactions feel more natural.
- Smoother search updates: ChatGPT Voice will use a wider variety of short phrases to signal that it is searching.
- Gradual availability: The new options may vary by region and account while the rollout continues.
What Is GPT-Live-1.1?
GPT-Live-1.1 is the latest version of OpenAI’s GPT-Live family of voice models, which powers real-time spoken conversations in ChatGPT’s Live voice experience.
Unlike a conventional text chatbot that waits for a message and then produces a written answer, a real-time voice model must manage speech, pauses, interruptions and the timing of responses. The system needs to determine when a person has finished speaking, when to respond and when it should continue listening.
OpenAI designed its Live models to listen and speak in a more continuous way. This helps conversations feel less like a sequence of recorded questions and answers and more like an interactive discussion.
The GPT-Live-1.1 update builds on that approach. Its most visible changes involve more natural accents, expanded voice choices and improvements to how the assistant communicates while carrying out tasks.
OpenAI has also introduced GPT-Live-1.1 Quip, a smaller model intended for users who receive the lighter version of the Live experience. The model assignment depends on the applicable plan and product configuration.
The update is being rolled out across ChatGPT, with related availability in Codex and Dots also described in OpenAI’s developer announcements. API access is expected to follow, although a specific public release date for developers has not been confirmed in the available announcement.
ChatGPT Voice Gets 13 New Voice Options
The expanded voice lineup is one of the most noticeable parts of the release. OpenAI is adding options across several languages, giving users more ways to choose how spoken responses sound.
| Language or variety | New voice options |
|---|---|
| English | Bubbie |
| French | Two new voices |
| German | Two new voices |
| Hindi | Neer and Leher |
| Japanese | Two new voices |
| Korean | Two new voices |
| Mexican Spanish | Two new voices |
The additions total 13 new voice options: two for each of six language varieties, plus Bubbie in English.
The expansion matters because voice selection is not simply a cosmetic feature. A voice that sounds natural to one user may feel unfamiliar to another, especially when pronunciation, rhythm and accent influence how easy a spoken answer is to understand.
Users may also prefer different voices for different purposes. Someone practising a language may want clear pronunciation, while a person using voice mode for everyday tasks may prioritise a conversational delivery.
OpenAI notes that the voices available to a user can vary by region and account. Therefore, the full lineup may not be visible to everyone at the same time.
Why the Update Matters for Hindi-Speaking Users
The inclusion of Neer and Leher makes the rollout particularly relevant to India, where many people communicate more comfortably through spoken language than through long written prompts.
Voice interfaces can help users ask questions while travelling, working, cooking or carrying out other activities that make typing inconvenient. They can also help people who are more comfortable speaking than writing in English.
OpenAI has been increasing its focus on multilingual and voice-based experiences in India. Company executives have described Hindi-tuned voice capabilities as part of the effort to make conversations feel more natural for local users.
A more natural-sounding voice, however, is only one part of useful multilingual AI. The system also needs to understand what a user says, interpret local expressions and respond accurately in context.
India’s linguistic diversity adds another challenge. A person may use Hindi alongside English, switch between languages in the same sentence or speak with a regional accent. Improving the spoken output can make interactions more comfortable, but it does not guarantee perfect recognition or pronunciation in every situation.
The new Hindi voices therefore represent a step in the development of voice interfaces, rather than proof that every regional-language interaction has been solved.
More Natural Accents and Conversational Delivery
OpenAI says GPT-Live-1.1 improves the naturalness of accents across languages. The objective is to make spoken responses feel more appropriate to the language being used instead of relying on a delivery that sounds overly uniform.
Natural speech involves several elements. Pronunciation is important, but so are intonation, rhythm, pauses and emphasis. A sentence can contain the correct words and still sound unnatural if its stress patterns do not fit the language.
These details matter when users rely on voice mode for explanations, language practice, brainstorming or longer discussions. A more natural delivery may make it easier to follow an answer and sustain a conversation.
The update also gives users more choices about the voice they hear. That can improve personalisation, although voice preference remains subjective and different users may prefer different speaking styles.
OpenAI has not presented the release as a guarantee that all speech will be flawless. Voice models can still misunderstand spoken words, mispronounce names or produce incorrect information. Users should continue to verify important answers, especially when the conversation involves health, finances or other consequential decisions.
ChatGPT Voice Will Use More Varied Search Phrases
Another change concerns the short spoken updates ChatGPT gives while searching for information.
Previously, a voice assistant could repeatedly use similar phrases to signal that it was checking something. GPT-Live-1.1 is designed to use a wider range of short phrases during searches, according to OpenAI’s release notes.
This is a small change, but it affects how the assistant communicates during a task. When a voice assistant pauses to retrieve information, a brief spoken update can reassure the user that the request is still being handled.
More varied wording may make these moments feel less repetitive. It can also make the conversation sound less scripted, particularly when users ask several questions that require web searches.
The improvement does not mean every search will be faster or that all results will be more accurate. It is primarily a change to the way the assistant communicates while searching, rather than a promise of improved web-search performance in every case.
GPT-Live-1.1 and GPT-Live-1.1 Quip: What Is the Difference?
OpenAI is rolling out two related models: GPT-Live-1.1 and GPT-Live-1.1 Quip.
The full GPT-Live-1.1 model is the default voice model for Plus and higher-tier users, including use in Codex and Dots, according to OpenAI’s system card. GPT-Live-1.1 Quip is the successor to the smaller GPT-Live-1 mini model and is the default for other eligible users.
The distinction allows OpenAI to offer different voice experiences across subscription tiers and product configurations. However, the company has not announced that the update changes every user’s daily usage allowance.
| Model | General role |
|---|---|
| GPT-Live-1.1 | Full Live voice model for Plus and higher-tier users |
| GPT-Live-1.1 Quip | Smaller Live voice model for other eligible users |
Availability and usage limits can depend on a user’s plan, account and current product settings. Users should check the latest Voice FAQ and their account’s settings rather than assume that the model update automatically changes their access limits.
How to Try the New ChatGPT Voices
The rollout is gradual, so users may need to wait before the new voices appear in their accounts.
To check for the update:
- Open ChatGPT on the web or in the mobile app.
- Open Settings.
- Select Voice.
- Review the available voice options and select one.
- Start a Live voice conversation to try the selected voice.
If the new options are not visible, it does not necessarily mean something is wrong with the account. OpenAI says availability can vary by region and account, and the rollout may not have reached every user yet.
Users should also make sure they are using an up-to-date version of the mobile app when checking for changes.
What the Update Means for the Future of AI Voice
The release reflects a broader shift in how people interact with AI assistants. Text remains useful for detailed instructions, editing and reviewing information, but spoken conversations can make AI easier to use in situations where typing is inconvenient.
As voice models improve, the challenge is to combine natural speech with reliable understanding, useful answers and predictable behaviour. A pleasant voice alone is not enough if the assistant misunderstands a request or provides inaccurate information.
Multilingual performance is another important part of that challenge. People often want to interact with technology in the language they use at home, at work or with their communities. More natural voice options may lower the barrier to using AI, especially for people who are less comfortable with English-language interfaces.
The new model is one step in that direction. Its practical value will depend on how well it performs across real conversations, accents and languages, and whether users find the experience more helpful than previous versions.
The Bigger Picture
GPT-Live-1.1 shows that OpenAI is continuing to invest in the spoken side of ChatGPT, not just in text-based reasoning. The new voice options, more natural accents and less repetitive search updates target the details that shape everyday interactions. Hindi support is particularly relevant as the company seeks to make AI more accessible to a wider range of users in India.
The update also illustrates the difference between product improvements and guaranteed outcomes. More voice choices and smoother delivery can make conversations more comfortable, but they do not eliminate errors or ensure identical availability across all accounts. The longer-term test will be whether these changes help users communicate more naturally while preserving the accuracy and reliability they expect from an AI assistant.
Looking Ahead
OpenAI is gradually rolling out GPT-Live-1.1 and GPT-Live-1.1 Quip across eligible experiences. Users can check Settings under Voice for new options, while developers will need to wait for the announced API availability. The timing of access may differ by account and region, and OpenAI has not specified a universal completion date for the rollout.
Future improvements to voice AI will likely be judged on more than the number of available voices. Pronunciation, language understanding, interruption handling, responsiveness and factual reliability will all matter. For now, GPT-Live-1.1 represents an incremental but meaningful update to the way ChatGPT speaks and manages live conversations.
Frequently Asked Questions
1. What is GPT-Live-1.1?
GPT-Live-1.1 is OpenAI’s updated voice model for ChatGPT Live. It focuses on more natural accents, expanded voice options and smoother spoken updates during searches.
2. How many new voices is OpenAI adding?
OpenAI is introducing 13 new voices across French, German, Hindi, Japanese, Korean and Mexican Spanish, plus Bubbie in English.
3. What are the new Hindi voices called?
The new Hindi voice options are Neer and Leher.
4. How can I access GPT-Live-1.1?
Open ChatGPT, go to Settings, select Voice and review the available options. The rollout is gradual, so the new voices may not appear immediately.
5. Is GPT-Live-1.1 available to everyone?
OpenAI is rolling out the update across eligible ChatGPT experiences, but the model and voices available can vary by plan, region and account. GPT-Live-1.1 Quip is the smaller model for users assigned the lighter Live experience.
6. Does GPT-Live-1.1 guarantee more accurate answers?
No. The update focuses on voice delivery, accents and conversational flow. Like other AI models, it can still make mistakes, so important information should be checked.
Get the day’s top stories in your inbox
One concise email. No spam, unsubscribe anytime.
