OpenAI brings ChatGPT Voice to desktop app, enabling voice-controlled AI agents and computer tasks

OpenAI on Thursday announced that its ChatGPT Voice feature is now available in the ChatGPT desktop app, allowing users to control AI agents and perform computer tasks using spoken commands. The update brings the voice model family called ChatGPT-Live, launched earlier this month, to the desktop environment, and works with both ChatGPT Work and Codex.

The move extends the conversational voice mode that was previously limited to smartphones. On desktop, the feature is more capable: users can dictate multi-step commands, such as creating a new thread, making a pull request, and finding the root cause of a bug, all in one sentence. OpenAI also demonstrated the ability to use computer vision via Appshots on macOS, letting the app read screen content including alt-text.

How ChatGPT Voice works on desktop

ChatGPT Voice on desktop uses the GPT-Live model, which is designed to simultaneously speak, listen, and coordinate tasks within the app. Users can issue commands that involve multiple steps, and the assistant responds when it needs additional input. The voice mode can also tap computer use skills to look up websites and applications, effectively acting as a hands-free interface for everyday computing tasks.

OpenAI said the feature is rolling out globally today. On macOS, users can allow the app to access on-screen content through Appshots, enabling context-aware actions. For example, a developer could ask ChatGPT to analyze a bug shown on screen and then fix it by speaking instructions.

Integration with ChatGPT Work and Codex

The voice mode works directly with ChatGPT Work, OpenAI’s workspace for collaborative AI tasks, and Codex, its code-generation environment. Users can direct multiple agents running in these platforms using voice alone. This opens up hands-free workflows for developers, project managers, and content creators who need to manage complex, multi-step processes without typing.

OpenAI also noted that users can already access ChatGPT Voice in Codex from the iOS app through remote control, but the desktop version provides a more integrated experience. The company’s demo video showed a developer saying a single command that triggered a new thread, a pull request, and a bug investigation — all without touching the keyboard.

Comparison with Anthropic’s Claude voice mode

The announcement comes shortly after Anthropic updated its voice mode for Claude, which can tap the company’s Opus, Sonnet, and Haiku models. Claude’s voice mode focuses on completing tasks within apps like Gmail, Calendar, Slack, Notion, and Canva. While both products aim to make AI interaction more natural, OpenAI’s desktop version leans into computer control and agent orchestration, while Anthropic emphasises integration with productivity tools.

For Indian users, the availability of ChatGPT Voice on desktop means that developers, digital marketers, and AI researchers can now incorporate voice commands into their workflows without switching to a phone. This is particularly relevant for power users who rely on Codex for code generation or ChatGPT Work for managing AI agents.

What this means for Indian AI users

The rollout of ChatGPT Voice on desktop aligns with the growing adoption of AI agents in Indian tech startups, digital marketing agencies, and software development teams. Voice control reduces the friction of typing long prompts and allows multitasking. For example, a digital marketer could use voice to ask ChatGPT to analyse SEO data, draft a social media post, and schedule it — all in one conversation.

However, the feature requires a stable internet connection and a compatible desktop app. OpenAI has not yet specified whether it will be available on all desktop platforms or only on macOS and Windows. The company has also not disclosed pricing for the voice mode beyond the existing ChatGPT Plus subscription.

Datos clave

Key detail	Information
Feature	ChatGPT Voice on desktop app
Model	GPT-Live (new voice model)
Integration	ChatGPT Work, Codex, computer vision (macOS)
Availability	Rolling out globally from July 24, 2026

What to watch next

Indian users should check whether the feature is available in their region, as global rollouts often take a few days. The ability to control multiple agents with voice could become a standard workflow for AI-first teams. Developers using Codex should test the feature for hands-free debugging and code review. It remains to be seen whether OpenAI will extend voice control to third-party desktop apps or limit it to its own ecosystem.

Source: TechCrunch — OpenAI’s new voice mode makes it to the ChatGPT desktop app (https://techcrunch.com/2026/07/24/openais-new-voice-mode-makes-it-to-the-chatgpt-desktop-app/)