Two weeks after releasing his more naturalistic GPT-Live audio AI model With full-duplex capabilities (listen and talk at the same time), OpenAI incorporates them directly into developer workflows.
The company announced that GPT-Live now powers the ChatGPT desktop app on macOS and Windows, integrating directly with agent systems like Codex and ChatGPT Work (which are standalone experiences available in the ChatGPT desktop app).
When OpenAI initially released GPT-Live on July 8, 2026, it introduced a continuous audio model capable of listening and speaking simultaneously, removing the rigidity of turn-taking and delegating complex reasoning to background models like GPT-5.5.
Today’s release extends that conversational layer to technical tasks, allowing software engineers to orchestrate multi-threaded coding jobs, review pull requests, and debug applications using natural voice commands.
As such, it could usher in a new era of "hands free" software development and even live, in-person group coding parties to more than 10 million weekly active users throughout the Codex and ChatGPT work. Codex, of course, is the name given to OpenAI’s coding-focused models and harnesses, but which the company expanded this year to a larger version. general productivity platform. An OpenAI spokesperson told VentureBeat that this is the first time it has been voice-activated.
OpenAI published a promotional video showing some of their employees, Codex developer experience engineer Jason Liu and Codex technical staff member Guinness Chen, speaking in the same ChatGPT desktop app session in the same room, each giving different instructions and conversing with the same model.
New abilities unlocked
In essence, this integration is based on decoupling the real-time voice layer from the underlying execution engines.
While GPT-Live maintains a fluid conversation, inserting natural verbal recognitions such as "I understand" without interrupting the user: moves heavy computational workloads to reasoning models in the background.
On macOS, the desktop app includes "Application Photos" and screen context functions, allowing ChatGPT Voice to analyze the front window along with local files, codebase structures, and active plugins.
This architecture creates a pair programming dynamic in which developers talk about problems conversationally while agents execute tasks asynchronously.
Instead of manually stopping coding sessions to type detailed instructions or switch windows, developers run the system hands-free.
The full-duplex engine dynamically decides when to talk, pause, or invoke tools, maintaining the conversation state even when background agents process complex code modifications.
Drive complex coding and builds with just your voice
The core operational capability of this update focuses on executing multiple tasks in the Codex and ChatGPT work environments.
Software engineers can launch multiple concurrent task threads from a single spoken message. For example, a developer preparing to release a feature can instruct the system to investigate an open authentication error, review a pending API migration pull request, and generate missing unit tests simultaneously.
The desktop app coordinates these actions across disparate contexts, tracking issues across Slack conversations, GitHub repositories, and local codebases.
Developers can also verbally convert design mockups into working code, dividing tasks between the frontend, backend, and test layers.
With support for multi-folder projects (build 26.715) and remote execution via iOS, engineers can check task progress, respond to agent requests, and reroute active jobs without switching apps or managing individual processes line by line.
proprietary license
The voice-enabled desktop version of OpenAI operates under a proprietary commercial business model. Access is restricted to paid subscribers on Plus, Pro, Business, Enterprise and Education plans.
For individual developers and corporate engineering departments, this business structure means that model weights, speech processing pipelines, and agent state architectures remain completely closed.
Organizations cannot modify or host the underlying systems themselves. Additionally, tasks initiated through ChatGPT Voice consume standard usage allocations directly from existing Codex and ChatGPT work plan quotas, and treat voice-triggered actions identically to standard agent workloads.
Community reactions
Developer communities immediately noticed the implications of bringing full-duplex continuous voice to standalone encoding workflows.
Reacting to the announcement of the release of build 26.715, which details voice integration and multi-folder project support, AI Insider journalist @ChrisGPT noted in X: "Today OpenAI will launch remote and voice guidance for the codex! One step closer to personal AGI".
Early technical feedback highlights widespread enthusiasm for orchestrating complex agent tasks hands-free, especially when moving away from the workstation or managing build processes remotely.
