Affiliate links on Android Authority may earn us a commission. Learn more.
ChatGPT leak reveals new Bidi 1 voice model that can listen and speak simultaneously
2 hours ago
- OpenAI is reportedly testing an unannounced bidirectional voice model called “GPT-Bidi-1.”
- Code references and early user tests show that the model can speak, hear, and listen simultaneously, handling mid-sentence interruptions naturally.
- The unannounced model has already started rolling out to a select group of app users, hinting at an official release window this week.
OpenAI is reportedly planning to turn ChatGPT into a superapp, with a major overhaul in the pipeline. The overhaul is said to focus on OpenAI’s Codex coding tool and agentic AI tools that can perform tasks for users. But there seems to be more in store, as a new bidirectional audio model named “GPT Bidi 1” has also been spotted, which would be a massive upgrade to ChatGPT’s conversational abilities.
Bidi is said to be shorthand for bidirectional design, which allows the assistant to speak, hear, and listen simultaneously. TestingCatalog spotted references to Bidi 1 last week, with internal code presenting it as a “major leap in intelligence,” and “the next generation of Voice.”
Bidi 1 is said to sit in the model selector under settings, besides the standard and advanced options. The voice bubble turns yellow once Bidi 1 is picked.
According to a recent report from TestingCatalog, the new model has already begun rolling out to a subset of ChatGPT app users, suggesting a possible release this week.
The model is said to offer small, natural acknowledgments, like an “okay,” when you pause or slow down, without cutting you off. It is also said to switch tasks on the fly: for example, ask it to count to ten, interrupt to reverse the count, and it adjusts immediately.
BREAKING 🔥: First tests of “Bidi 1”, an upcoming bidirectional voice model from OpenAI. This upgrade will arrive in ChatGPT and, potentially, in Codex soon as well.> Bidi 1 can speak over while you are talking and keep listening.
> Bidi 1 can switch between tasks back and… https://t.co/BwWhCKx3G0 pic.twitter.com/Fawc74kBym— 🚨 AI News | TestingCatalog (@testingcatalog) June 23, 2026
Perhaps one of the biggest changes would be that the model is said to hold the thread of the whole long conversation, rather than dropping earlier context, the weak point that has long dogged ChatGPT’s current voice stack. It also no longer jumps in during longer pauses.
The report further highlights that the release of Bidi 1 can be seen as a way for OpenAI to close the gap between its capable text models and its older voice layer. This is important because OpenAI is betting that speech will be the primary way most people access AI, rather than text.
OpenAI has not yet announced the Bidi 1 model, nor has it detailed GPT 5.6 yet. Hopefully, the company releases official details soon.
Thank you for being part of our community. Read our Comment Policy before posting.