OpenAI has released the API for GPT-Live-1, a new voice model designed to create more natural and controllable voice interactions. This model supports full-duplex communication, allowing AI agents to speak and listen simultaneously, and can delegate complex tasks like reasoning or tool use to other models. Developers can customize the AI's tone, pacing, and conversational style, with improvements in handling background noise and long sessions. GPT-Live-1 is already powering new voice modes in ChatGPT and offers a 30% performance increase over its predecessor, GPT-Realtime-2.1. AI
IMPACT Enables more natural, interactive, and customizable voice AI experiences, potentially accelerating adoption in applications like customer support and personal assistants.
RANK_REASON Frontier-lab model release with system card and API availability. [lever_c_demoted from frontier_release: ic=1 ai=1.0]
Read on Mastodon — sigmoid.social →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
