October 1, 2026
You can now talk to Lovable apps: voice conversations run on GPT-Live
A voice companion in an app can keep the conversation going while a separate AI agent carries out a task.
Lovable
@lovable
You can now add live voice conversations powered by GPT-Live from @OpenAI to your Lovable apps. @mgfeller built an app where you can talk to Rocky from Project Hail Mary. Love it, love it, love it.
· 3.8K views
In an app built with Lovable, you can talk to Rocky from Project Hail Mary. It is a demo of the new voice capability.
Lovable now lets you add live voice conversations powered by GPT-Live to your apps. GPT-Live listens and speaks simultaneously. It delegates searches and tasks to a separate agent while keeping the conversation going.
Conversation cost. According to OpenAI's documentation as of Oct 1, 2026, a gpt-live-1 session through the API costs $0.05 per minute, billed by the second. A ten-minute conversation costs $0.50, with the separate model and tools billed additionally.
Your own browser version. For direct integration through the API, OpenAI provides a ready-to-run WebRTC example. To get started:
1. Save the `server.mjs` and `index.html` files from the example. 2. Run `npm install openai express`. 3. Set `OPENAI_API_KEY` on the server.
The example requires Node.js 22.6 or later and runs over HTTPS or on localhost.
Direct API access to gpt-live-1 starts at Tier 1: Free is not supported, and concurrent session limits are 25 for Tier 1, 50 for Tier 2, 200 for Tier 3, 300 for Tier 4 and 500 for Tier 5.
Source
