Hello Alena Dauhalionak,
Welcome to Microsoft Q&A,
What you’re seeing is mostly due to two different UI generations and entry points for Voice Live in Microsoft Foundry:
- The YouTube tutorial uses a preview UX where the Agent playground had a direct checkbox: “Use voice live on this agent”.
- In the current public experience, voice is wired through the Voice Live / Speech playground, not via a checkbox inside the Agent playground. That’s why you don’t see that option even though you’re on the same model (gpt-5-chat).
Instead of a checkbox on the Agent itself, the current docs show Voice Live being used from the Speech / Voice Live playground, where you either:
- Use a model directly (for example,
gpt-5-chat), or - Attach an existing Foundry Agent created in the Agents playground.
- Go to your Microsoft Foundry project.
- In the top menu, select Build.
- On the left, select Models.
- Under AI Services, choose Azure Speech – Voice Live to open the Voice Live playground.
- Pick a scenario (for example, Casual chat) and a voice.
- Under Configuration → GenAI → Generative AI model, either:
- Select
gpt-5-chatdirectly, or - Use the option to select one of your Agents (if exposed for your tenant – this is how you “give an agent a voice” in the new flow).
- Select
- Click Start and talk using your mic; click End to stop.
Please Upvote and accept the answer if it helps!!