OpenAI has integrated its full-duplex GPT-Live voice model into the ChatGPT desktop application, enabling developers to orchestrate coding tasks in Codex and ChatGPT Work entirely through natural speech.
The update, announced July 23, lets software engineers review pull requests, debug applications, and manage multi-threaded coding jobs using voice commands. GPT-Live continuously listens and speaks without rigid turn-taking, passing heavy computational workloads to background reasoning models while maintaining fluid conversation.
In a promotional video, OpenAI showed multiple employees speaking to the same ChatGPT desktop session simultaneously, each issuing different instructions to the same model instance. The integration supports Appshots and screen context on macOS, allowing ChatGPT Voice to analyze the frontmost window, local files, codebase structures, and active plugins.
Developers can launch multiple concurrent task threads from a single spoken prompt. A user preparing to ship a feature can ask the system to investigate an authentication bug, review a pending API migration pull request, and generate missing unit tests simultaneously across Slack, GitHub, and local codebases.
The feature is available to paid subscribers across Plus, Pro, Business, Enterprise, and Education plans. OpenAI first launched GPT-Live on July 8 with full-duplex capabilities that eliminated the awkward pause-and-respond pattern of earlier voice interfaces.
