OpenAI has taken ChatGPT's voice mode to a new level. Now the user can not only ask questions, but also perform multi-step tasks by voice: create a document, work with connected services or perform actions in the browser. The updated feature is available in the web version and mobile devices, reports Tech-news. The new system is based on the GPT-Live-1 voice model. It can conduct the conversation in real time and listen while the user is speaking. And tasks that require more complex calculations and tools can be assigned by the system to other models and agents. For example, through Voice, a user can ask ChatGPT to prepare a schedule or presentation, work with connected applications, or perform certain actions in the browser. Some tasks require additional confirmation from the user on the screen. This change also expands the role of the Voice function. Before it was mainly a convenient way to communicate with ChatGPT, but now the voice interface is becoming a real task management tool. According to OpenAI, GPT-Live can continue voice dialogue even when complex work is in progress. That is, the user continues to communicate with the system without waiting for the task to be completed.