OpenAI has launched GPT-Live, a groundbreaking full-duplex voice model that enables ChatGPT to engage in truly natural, real-time conversations. The new model rolled out globally on July 8, 2026, across iOS, Android, and ChatGPT.com, marking a significant leap toward human-like voice interaction.
Two Tiers for Global Users
The launch includes two models: GPT-Live-1 and GPT-Live-1 mini. GPT-Live-1 becomes the default voice model for paid ChatGPT users on the Go, Plus, and Pro tiers, while GPT-Live-1 mini serves free-tier users. OpenAI plans to bring both models to the API, with developers able to sign up for notifications.
Full-duplex audio means the model can listen and speak simultaneously, eliminating the awkward pauses that characterized previous voice assistants. This enables interruptions, backchanneling, and the natural rhythm of human conversation—features that have long eluded AI voice systems.
Why Full-Duplex Matters
Traditional voice assistants operate in half-duplex mode: they listen, then process, then respond. This creates a mechanical feel that users have grown accustomed to accepting as the limitation of AI. GPT-Live’s full-duplex architecture fundamentally changes this dynamic.
The model processes audio streams continuously, maintaining context and understanding while generating responses. This approach mimics how human brains handle conversation—simultaneously listening, thinking, and preparing to speak rather than treating these as discrete, sequential operations.
Market Implications
The launch puts pressure on competitors including Google’s Gemini Live and Anthropic’s voice capabilities. With GPT-Live now defaulting to paid tiers, OpenAI signals that advanced voice interaction is transitioning from experimental feature to core product differentiator.
The API availability coming soon will enable developers to build applications leveraging the same conversational voice technology, potentially accelerating adoption across customer service, education, and accessibility tools.