Summary
- GPT-Live is an updated, advanced voice model for ChatGPT.
- GPT-Live aims to be more conversational while handling deep queries in the background for a seamless feel.
- A global rollout began on July 8 — users on paid plans will get GPT-Live-1, and Free users will get GPT-Live-1 mini, a slightly scaled-back version.
On July 8, OpenAI announced GPT-Live, an upgraded voice experience for ChatGPT. GPT-Live marks a major shift in ChatGPT’s conversational capabilities, with a more natural feel and smarter responses.
What is GPT-Live?
Next-gen conversational AI
GPT-Live is OpenAI’s next-generation voice model. It aims to “make talking with AI feel much more like having a real conversation.” GPT-Live is built on a “full-duplex architecture,” which means it can listen and speak simultaneously.
According to OpenAI, the model shows it’s paying attention by using phrases like “mhmm” or “yeah” and engaging in back-and-forth conversation. It should also be able to tell when you’re thinking and wait quietly for you to continue. The goal is “a voice experience that is refreshingly easy to talk to.”
OpenAI also claims GPT-Live is its “smartest voice model yet.” It can delegate to the latest frontier models when necessary (GPT-5.5 at launch), and it can also continue the conversation while it thinks in the background, so conversations flow more naturally.
Why GPT-Live is a big deal
Playing catch-up
This is a big step for ChatGPT’s voice assistant. In the past, we’ve noted that other LLMs like Claude have felt more natural. And while the platform has had voice models for a while now, it’s typically lagged behind competitors in this area, too — especially Gemini.
These new GPT-Live models should go a long way toward making ChatGPT more natural to use and help bring it up to par with Gemini Live and other competitors. Combined with features like Scheduled Tasks, ChatGPT could be a compelling alternative to Gemini.
OpenAI also claims that GPT-Live is built with enhanced safeguards that can either steer responses to safer territory or end the conversation entirely in high-risk situations. There are “additional protections to support teen users,” and the models are trained on age-appropriate behavior to tailor responses to the listener.
GPT-Live availability
Global rollout
GPT-Live began rolling out globally on July 8. It’s available across iOS, Android, and the web. Two models are being rolled out, depending on your subscription tier. Users on Go, Plus, and Pro will get access to GPT-Live-1, while Free users will get access to GPT-Live-1 mini (a similar but slightly scaled-down version). In both cases, GPT-Live will be the default model for ChatGPT Voice.
The main limitations at this time are in language fluency and video. GPT-Live is optimized for “some of the most popular languages in ChatGPT,” but OpenAI warns that certain languages may have a non-native accent or “gaps in fluency.” The company says it’s actively working on this.
As for video, GPT-Live does not currently support voice with video or screen sharing. Again, OpenAI says it’s working on this and will introduce these capabilities “soon.” In the meantime, if you need these features, legacy versions of ChatGPT Voice that support them are still available.
- OS
-
Android, iOS, Web
- Developer
-
OpenAI
- Price model
-
Free with optional subscription
ChatGPT is OpenAI’s flagship product, with powerful features. It was just updated with advanced, conversational voice capabilities.
