Introducing GPT‑Live

OpenAI has launched GPT-Live, a significant upgrade to the voice mode feature within the ChatGPT mobile application. This new model replaces the previous version, which was based on older GPT-4o technology and suffered from limited reasoning capabilities and an outdated knowledge cutoff.

A key feature of GPT-Live is its ability to delegate complex tasks to the more powerful GPT-5.5 model. When a user asks a question requiring web searches, deep reasoning, or intricate analysis, the system handles these requests in the background. This allows the voice conversation to continue uninterrupted while the more advanced model processes the information and returns the final answer to the user.

The system is designed to evolve, meaning OpenAI will automatically update the underlying frontier model as new versions are released. During early testing, users noted improvements in conversational flow, though some initial bugs occurred, such as the model inappropriately interrupting to laugh at non-humorous statements. These issues have since been addressed through adjustments to the model.

This update is significant because it transforms ChatGPT voice mode from a limited tool into a more capable brainstorming partner. By integrating real-time task delegation with a more advanced model, the system provides a more fluid and intelligent experience for users engaged in long-form verbal interactions.