OpenAI rolled out a major update to its ChatGPT desktop client on Thursday, adding a voice‑first interface that lets users talk to the app to direct AI agents and execute computer tasks. The new capability, called ChatGPT Voice, runs on the company’s recently unveiled ChatGPT‑Live family of voice models, which debuted earlier this month.
Unlike the earlier smartphone version, which focused on smoother conversation flow and better interruption handling, the desktop integration can carry out complex, multi‑step commands. In a demonstration, a developer asked the assistant to create a new discussion thread, open a pull request and locate the root cause of a bug—all with a single spoken instruction. The system responded by coordinating with both ChatGPT Work and Codex, showing that the voice mode can orchestrate multiple agents simultaneously.
ChatGPT Voice also taps into the app’s “computer use” skills, allowing it to browse websites, launch applications, and retrieve information from the user’s screen. On macOS, the feature leverages OpenAI’s Appshots technology, granting the assistant access to on‑screen content such as alt‑text descriptions. Users can therefore dictate intricate workflows that involve looking up data, filling forms, or manipulating files without touching a keyboard.
Availability and rollout
The update is being pushed to all desktop users worldwide starting today. OpenAI said the voice mode works with both the ChatGPT Work environment and the Codex developer platform, extending its reach beyond casual chat to professional coding and collaboration scenarios. While the voice interface is not yet built to execute actions on smartphones, the company noted that Codex users can still access Voice through the iOS app via remote desktop connections.
OpenAI shared a short video on its X account illustrating the new functionality, and a tweet highlighted the global rollout with a screenshot of the desktop interface listening for commands. The company did not disclose pricing changes, implying the feature is included for existing ChatGPT subscribers.
Anthropic, a rival AI firm, also announced an upgrade to its Claude voice mode, enabling the assistant to operate within Gmail, Calendar, Slack, Notion and Canva using its Opus, Sonnet and Haiku models. The parallel announcements suggest a growing emphasis on voice‑driven AI across the industry.
Dieser Artikel wurde mit Unterstützung von KI verfasst.
News Factory APP - agentische News für besseres SEO & AEO.