OpenAI has launched GPT-Live-1 in its API, a voice model that enables natural, full-duplex conversations and cuts interruptions by nearly 80%. The model, first introduced in ChatGPT, is now available for developers to build voice-enabled apps and business workflows.
TL;DR
- GPT-Live-1 enables natural, full-duplex voice conversations, reducing interruptions by 80% compared to turn-based systems.
- The model simplifies voice-agent architecture by handling listening and speaking in a single model, reducing latency and improving context retention.
- GPT-Live-1 is available in the API today, with pricing starting at $0.05 per minute for the front-end voice layer.
What happened
GPT-Live-1, initially introduced in ChatGPT, is now available in the OpenAI API, allowing developers to build voice-enabled applications. The model supports full-duplex conversations, enabling simultaneous listening and speaking, which significantly improves the natural flow of interactions.
Key features of GPT-Live-1 include interruption handling, reasoning and tool calling delegation, customizable tone and pace, silent context management, long-session reliability, and telephony support. These features allow developers to create more engaging and efficient voice experiences for users.
OpenAI has also expanded voice options, offering a broader selection of accents, dialects, and languages. The company plans to continue expanding voice options and language availability in the coming months.
Why it matters
For developers, GPT-Live-1 simplifies voice-agent architecture by handling listening and speaking in a single model. This reduces latency and improves context retention, making it easier to build natural and engaging voice experiences. Developers can pair GPT-Live-1 with other models like Luna or Astra to match reasoning depth, speed, and cost to each task.
For startups and investors, the launch of GPT-Live-1 represents a significant advancement in voice technology. The model's ability to handle full-duplex conversations and reduce interruptions by 80% opens up new possibilities for voice-enabled applications and business workflows. This could lead to increased investment and innovation in the voice technology sector.
However, the model's performance may vary depending on the backend model and agent harness used. Developers should carefully consider their choices to ensure optimal performance and user experience.
Key facts
- GPT-Live-1 reduces interruptions in voice agents by nearly 80% compared to turn-based systems.
- The model improves Full Duplex Bench performance by 30 percentage points over GPT-Realtime-2.1.
- GPT-Live-1 is available in the API today, with pricing starting at $0.05 per minute for the front-end voice layer.
- The model supports custom voices, allowing developers to choose voices that fit their product and sound natural to users.
- GPT-Live-1 can delegate reasoning and tool calls to a backend text model like GPT-6 Astra or a third-party model.
- The model offers strong alphanumeric understanding and supports keyword biasing.
- GPT-Live-1 natively provides ASR transcripts and response text.
- OpenAI plans to expand voice options and language availability over the coming months.
Context
The launch of GPT-Live-1 is part of a broader trend in the AI industry towards more natural and engaging voice experiences. As voice technology continues to evolve, we can expect to see more innovations in this space, with a focus on improving the natural flow of conversations and reducing interruptions.
For developers, the availability of GPT-Live-1 in the API represents a significant opportunity to build more advanced voice-enabled applications. The model's ability to handle full-duplex conversations and reduce interruptions makes it a powerful tool for creating engaging and efficient voice experiences.
For startups and investors, the launch of GPT-Live-1 highlights the growing importance of voice technology in the AI industry. As more companies look to incorporate voice-enabled features into their products and services, we can expect to see increased investment and innovation in this space.
