OpenAI Upgrades ChatGPT Voice With GPT-6 Astra Models

OpenAI has rolled out a significant upgrade to its globally available ChatGPT Voice feature, transitioning the system to run on its latest GPT-6 Astra, Sol, and Luna models. This update introduces direct integration with external plugins for the first time, enabling the voice assistant to interact with essential workplace tools such as email, calendar applications, and Slack. Users can access these new capabilities immediately by updating to the latest version of the ChatGPT mobile or web application.

The integration allows users to execute complex workflows entirely through spoken commands. Demonstrated capabilities include retrieving calendar events, drafting and sending emails, and identifying or canceling duplicate charges within connected financial applications. In more advanced scenarios, the voice assistant can even assist in building a functional website complete with an integrated checkout page. Additionally, the update extends to ChatGPT Work on both web and mobile platforms, where professionals can dictate and generate documents, presentations, and spreadsheets.

For developers and industry practitioners, this release represents a shift from passive conversational AI to active, agentic workflows. By bridging the gap between natural language processing and third-party software APIs, OpenAI is moving closer to the concept of an omnipresent digital assistant. CEO Sam Altman has long championed this direction, referencing the AI assistant from the science fiction film Her when the initial version of ChatGPT Voice debuted back in 2024.

While these hands-free capabilities promise to streamline daily productivity and alter how humans interact with computers, they also introduce notable security and operational risks. Granting an AI voice model direct read-and-write access to sensitive communication channels like Slack and email requires robust guardrails to prevent accidental data exposure or unauthorized actions. Practitioners will need to carefully evaluate these security implications as they integrate these voice-driven agents into their daily professional pipelines.