GPT-Live-1.1 and Quip: What's New in ChatGPT Voice
More natural accents, new voices, and clearer search cues. How the models, plan limits, and developer access differ.
Eigent
OpenAI began rolling out GPT-Live-1.1 and GPT-Live-1.1 Quip on October 15, 2026. The update focuses on more natural multilingual accents, additional voices, and more varied phrases while ChatGPT searches. Here is what changes for users and what remains unproven.
What is GPT-Live-1.1?
GPT-Live-1.1 succeeds GPT-Live-1, which launched in July 2026. Both are full-duplex voice models: they can listen and speak at the same time, allowing interruptions and changes of pace. The update refines that conversation rather than introducing full duplex for the first time. OpenAI
What is GPT-Live-1.1 Quip?
Quip succeeds GPT-Live-1 mini. OpenAI describes it as the smaller model, with the full version serving Plus and higher tiers, including Codex and dots. Quip is a model name; selecting a different voice does not select a different model. The launch notes do not provide a general quality ranking between the two. OpenAI
New voices and search cues
The launch names new options for six languages plus Bubbie in English. These are the 13 voices associated with that announcement; the current voice catalog also lists Brazilian Portuguese options, Viola and Rio. Available choices can vary by region and account.
| Language | Voices |
|---|---|
| English | Bubbie |
| French | Aube, Rivage |
| German | Sonna, Miko |
| Hindi | Leher, Neer |
| Japanese | Himari, Rindō |
| Korean | Haetsal, Hanul |
| Mexican Spanish | Tinta, Arrullo |
Live also uses a wider variety of short phrases during searches. That may make a wait feel less repetitive; it does not establish faster search, better transcription, or more accurate answers. OpenAI
Which model and limits apply?
The following consumer allowances apply to Voice in Chat, not every voice surface. They were checked on October 16, 2026.
| Plan | Voice in Chat, rolling 24 hours |
|---|---|
| Free | Limited Quip access; limits can change |
| Go | 3 hours with Quip |
| Plus | 3 hours with GPT-Live-1.1 |
| Pro ($100/month) | 15 hours with GPT-Live-1.1 |
| Pro ($200/month) | Unlimited GPT-Live-1.1, subject to abuse guardrails |
Business and institutional plans have separate allowances or metered pricing. Voice in Work on web and mobile follows the plan’s Voice limits and also counts toward Work usage. Desktop Voice has separate pricing and limits. Check the official help page for your exact plan rather than applying this table to a desktop session. OpenAI
How to try the update
Open Settings > Voice, choose Live where available, then choose a voice. Test a short conversation in the language you use, interrupt a response, and ask a question that needs search. Compare pronunciation and task accuracy separately. Availability depends on rollout, region, app version, and workspace settings. Live does not support video or screen sharing; eligible mobile users can use Advanced for those features. OpenAI
What the evaluations establish
The system card reports no statistically significant safety regressions on its production-prompt evaluations. On targeted synthetic prompts, GPT-Live-1.1 regressed in sexual content, mental health, emotional reliance, and gore categories. OpenAI says it plans improvements. These difficult test sets do not measure the frequency of failures in ordinary traffic, and the results do not establish improved speech recognition or factual accuracy. OpenAI
What developers can use
The API model page currently documents gpt-live-1. That is evidence of GPT-Live API access, not confirmation that GPT-Live-1.1 or Quip is available under the same identifier. Verify the supported model before building against the newer version. Keep the voice interaction, backend reasoning, and permission checks distinct. OpenAI
Connect conversation to a workflow
ChatGPT Voice can use connected apps and coordinate tasks where supported. If you are evaluating an open-source desktop workspace for multi-agent work, consider Eigent. Local execution does not mean all data stays on-device: cloud models and connectors can send data to their providers. Review the AI office agents comparison, then download Eigent and test one bounded workflow. ChatGPT Voice Eigent AI office agents Eigent
Recent Posts
Muse Spark Open: Meta's Open-Weight Model for Coding and Agents, Explained
Muse Spark Open is Meta's 816B open-weight model for coding and agents. See its benchmarks, hardware needs, license terms, and how it compares to rivals.
Eigent
NVIDIA Nemotron 3.5 Super VL: What's New, Benchmarks, and How to Run It
NVIDIA Nemotron 3.5 Super VL explained: the open 120B multimodal model's benchmarks, architecture, licensing, and how to run it locally for AI agents.
Eigent
Eigent Release Notes v1.0.6: Models, Automations & More Reliable Sessions
Eigent v1.0.6 brings redesigned Models settings and Automations, a new Profile step, and fixes for Session recovery, approvals, drafts and file previews.
Douglas Lai
