GPT-Live

GPT-Live is OpenAI's full-duplex voice model designed for real-time conversations that listen and speak simultaneously. It supports interruption handling, delegates background tasks to frontier models, and offers 9 distinct voices. The platform also includes live translation across 70+ languages and visual information cards, making it ideal for natural, fluid dialogue.

The platform offers a Pro-Monthly plan for $31.50 USD, which includes 2100 credits per month, high-quality AI image generation, and powerful AI video generation features. A Max-Monthly plan is available for $60 USD with 4500 credits, while a Basic-Monthly option costs $15.96 USD with 840 credits. All plans allow cancellation anytime and are valid for one month.

GPT-Live is specifically designed for users who need natural, real-time AI interactions, such as those practicing languages, commuting, or engaging in storytelling. It is also suitable for individuals seeking live translation across 70+ languages or those who require background task delegation for complex searches and deep reasoning while maintaining conversation flow.

The model processes input and output simultaneously, making interaction decisions multiple times per second to ensure fluid, natural back-and-forth communication. It filters background noise to focus on the user's voice even in busy environments and uses natural filler words to make conversations feel more human and responsive.

OpenAI announced on November 10, 2026, that the Standard and Advanced voice models would be retired on December 11, 2026. This change consolidates all voice conversation experiences into GPT-Live, which was launched in July of the same year, making it the sole model powering voice interactions.

The platform integrates several AI video generators, including Seedance 2.0, Gemini Omni Video, Kling 3.0, and MiniMax H3. Additionally, it provides free AI image generators like Nano Banana 2 and Grok Image Generator, along with tools for creating 3D science videos and AI video effects.

Users can access GPT-Live across multiple platforms, including iOS, Android, and ChatGPT.com. The service also offers a free tier for AI video and image generation, allowing users to create content without immediate cost for basic features.

The system supports interruption handling, allowing users to pause, redirect, or interrupt mid-sentence with natural acknowledgments like 'mm-hmm' and 'got it'. This feature ensures the conversation remains fluid and responsive to the user's immediate needs.

GPT-Live delegates complex tasks such as deep reasoning and search to frontier models like GPT-5.5 at different levels—Instant, Medium, and High. This ensures that answers are both fast and intelligent while keeping the conversation flow uninterrupted.

The platform offers live translation capabilities across 70+ languages, enabling real-time speech-to-speech translation. It also provides visual information cards and various AI effects tools to enhance user interaction and content creation.

The platform includes a variety of AI video models such as FLUX 3, Wan Dancer, Kimi K3, and PixVerse V6. These tools allow users to generate videos from text or images, creating diverse visual content for different needs.

The platform offers a MiniMax H3 video generator and a MiniMax H3 prompt tool, alongside a MiniMax H3 to video converter. These features are part of the comprehensive suite of AI-powered tools available for users.

The platform provides a free AI video generator and a free AI image generator, allowing users to create content without cost for basic tasks. These tools are part of the broader suite of AI effects and video generation capabilities.

You can reach GPT-Live by e-mail at support@seedance.ai.

The official website of GPT-Live is gpt-live.live.