Artificial intelligence assistants have made a new breakthrough in desktop interaction experiences. On Thursday local time, OpenAI announced an important update to its ChatGPT desktop application, officially launching support for ChatGPT Voice. With this new feature, users can now directly converse with ChatGPT through voice and control the AI agent to independently perform various complex tasks on their computers.

According to the information, this new feature is based on the newly launched ChatGPT-Live series of speech models by OpenAI earlier this month. ChatGPT Voice currently fully supports ChatGPT Work and Codex, and also has strong computer operation capabilities, helping users smoothly access various websites and applications. Especially on macOS systems, combined with the Appshots feature, users can even authorize the AI to access all content on the screen and alternative text (alt-text).

Compared to the previous mobile voice functions that mainly focused on providing a smooth conversation experience and optimizing interruption handling, this desktop update has achieved a qualitative leap in execution capability. Users can issue compound commands with simple voice commands and interactively respond when ChatGPT needs to supplement information or conduct a second confirmation. In the official demonstration, a developer successfully guided the AI to create a new code thread, submit a Pull Request, and quickly locate the root cause of a Bug with just one voice command. Additionally, OpenAI also revealed that users can use the remote access function of the iOS app to call ChatGPT Voice in Codex as well.

At the same time, industry competitors are also investing heavily in the voice interaction sector. Anthropic has also introduced a new voice mode for its Claude, which can call its Opus, Sonnet, and Haiku models to efficiently handle various daily work tasks in mainstream office applications such as Gmail, Calendar, Slack, Notion, and Canva.