Bỏ qua đến nội dung chính
Back to home
AI Tech 2 min read

OpenAI Integrates GPT-Live Voice Tech into Desktop App for Developers

OpenAI has brought GPT-Live's full-duplex voice control to its desktop app, enabling developers to orchestrate multi-agent workflows on Codex and ChatGPT Work.

Tier 1 · sources 68% confidence Reviewed
📚 Aggregated from 2 sources X — @sama VentureBeat – AI

OpenAI has officially released its full-duplex voice control feature powered by the GPT-Live model on its desktop application for macOS and Windows. This update allows users, particularly developers, to directly orchestrate agent systems like Codex and ChatGPT Work using voice commands instead of manual typing.

Key Details

According to posts by OpenAI on X and VentureBeat, this technology was rolled out globally on July 23, 2026, just two weeks after the GPT-Live audio model's debut on July 8, 2026. With the latest update (build 26.715), developers can now perform complex coding tasks—such as reviewing pull requests, debugging applications, or running parallel workflows—hands-free. OpenAI also demonstrated group interaction capabilities, where multiple engineers collaborated and issued commands to a single session on the desktop app without interference.

Technical Architecture

Architecturally, the system decouples the real-time voice processing layer from the underlying logic execution engines. While GPT-Live handles the continuous, two-way dialogue via its full-duplex capabilities (simultaneous listening and speaking), heavy computation and complex reasoning are offloaded to background foundational models (such as GPT-5.5). On macOS, the application also integrates an 'Appshots' feature and screen-context capture, enabling the voice assistant to directly analyze the active window, internal source code structures, and local files to deliver highly accurate responses.

Expert Opinions & Insights

The launch has received positive feedback from the developer community. Tech analyst @ChrisGPT noted on X that bringing remote guidance and voice control to Codex is 'one step closer to personal artificial general intelligence (AGI)'. However, as this is a proprietary commercial product of OpenAI, users must subscribe to Plus, Pro, Business, or Enterprise tiers to access it. Furthermore, voice-based execution will still consume standard account usage quotas, meaning enterprises must carefully evaluate operational costs.

Impact & Future Outlook

The commercialization of this hands-free programming tool promises to usher in a new workflow, reducing cognitive fatigue for developers who otherwise constantly switch between application windows. With Codex's weekly active user base already surpassing 5 million, OpenAI's move not only solidifies its leadership in AI-assisted development but also reshapes human-computer interaction through autonomous multi-agent systems in the near future.