OpenAI has expanded its latest voice capabilities to the ChatGPT desktop application, enabling users to interact with AI agents through natural speech while managing tasks across ChatGPT Work and Codex. The update extends the company's new GPT-Live-powered voice experience beyond mobile devices, marking another step toward voice becoming a primary interface for AI-assisted productivity.
The rollout introduces ChatGPT Voice to the desktop app on both macOS and Windows, allowing users to start conversations, assign tasks, monitor progress and coordinate AI-powered workflows using spoken commands. Unlike traditional voice assistants that primarily answer questions, the new experience is designed to control ongoing work across multiple AI agents without requiring users to switch between windows or type commands.
The desktop voice experience is powered by GPT-Live, OpenAI's latest family of conversational voice models introduced earlier this month. GPT-Live supports full duplex conversations, allowing the AI to listen and speak simultaneously, making interactions feel more natural by reducing interruptions and enabling users to interject mid-conversation. The models also integrate with OpenAI's latest reasoning systems and can access tools such as web search while maintaining an ongoing dialogue.
OpenAI said the new voice feature works across Chat, Work and Codex inside the desktop application. Users can verbally instruct the assistant to initiate tasks, check the status of AI agents, steer ongoing projects and coordinate multiple workflows from a single conversation. The feature also integrates with OpenAI's computer-use capabilities, allowing eligible users to control supported applications and perform desktop-based tasks using voice.
Mac users receive additional functionality through Appshots, which enables ChatGPT Voice to reference content visible on the active screen with user permission. This allows the assistant to better understand on-screen context when responding to queries or helping complete tasks. The feature is designed to improve productivity across document editing, coding, research and application workflows.
The launch follows OpenAI's broader effort to position voice as a more capable interface for AI systems. Earlier this month, the company unveiled GPT-Live-1 and GPT-Live-1 mini, describing them as more natural conversational models capable of longer interactions, better turn-taking and improved context handling compared with the previous Advanced Voice Mode. OpenAI has also said the models can remain silent while listening, allowing users to speak without being interrupted before responding.
The desktop rollout is available across paid ChatGPT plans, including Plus, Pro, Business, Enterprise and Edu, with availability expanding globally in phases. OpenAI has also introduced support for paired iPhone remote access, allowing users to continue desktop voice sessions through their mobile devices. Business customers can additionally use Voice in Work and Codex to coordinate enterprise workflows within their workspaces.
The update reflects increasing competition among AI companies to make conversational interfaces central to everyday computing. Technology firms are investing heavily in voice-first AI experiences that extend beyond answering questions to performing multi-step tasks, coordinating AI agents and interacting directly with applications. As enterprises and consumers adopt more agentic AI workflows, voice is emerging as an increasingly important interface for productivity and software interaction.
The latest desktop enhancement also aligns with OpenAI's broader strategy of integrating conversational AI more deeply across its ecosystem. By bringing advanced voice capabilities to desktop environments, the company is expanding how users interact with ChatGPT while supporting more complex workflows that combine natural conversation with AI-powered task execution.