← All models

GPT-Live-1

Published Publisher OpenAIgpt-live-1

Overview

GPT-Live-1 is OpenAI’s full-duplex voice model for natural, real-time conversations. It can listen and speak simultaneously, handle interruptions and pauses, and delegate deeper reasoning or tool use to a backend model while the conversation continues. (openai.com)

Capabilities and uses

  • Supports text and audio input and output, streaming, and function calling. It can provide speech transcripts and response text. (developers.openai.com)
  • Supports real-time translation and configurable voice tone, pace, and conversational style. Potential uses include language tutoring, customer support, phone-based agents, and other voice-enabled workflows. (openai.com)
  • Supported generated audio includes SynthID watermarking. (openai.com)

Limitations

  • Image and video input are not supported. Structured outputs, fine-tuning, and predicted outputs are not supported. The documented knowledge cutoff is July 31, 2025. (developers.openai.com)
  • More complex reasoning and actions can depend on the configured backend model and tools; their capabilities and safety characteristics apply to delegated work. OpenAI’s safety evaluations also report some category-specific regressions, so outputs should not be assumed infallible. (openai.com)

Release and availability

OpenAI introduced GPT-Live-1 in ChatGPT on July 8, 2026. The API release followed on September 10, 2026. (openai.com)

API pricing

As checked on November 16, 2026, the GPT-Live-1 front-end voice session costs USD 0.05 per minute, billed by the second without rounding up to a full minute. This is a session-duration price, not separate input- and output-token rates. Backend model and tool usage is billed separately under the applicable model and tool prices. (platform.openai.com)

Official sources