.NET: Bug: Magentic orchestration stalls when using newer OpenAI models than gpt-4.1

[Trojel]

Describe the bug
Magentic orchestration only works with Azure foundry gpt-4.1 or gpt-4o models. If the magentic manager and/or the agents use gpt-5 or any newer OpenAI model than gpt-4.1 the orchestration stalls and no agents are called by the manager

I have setup console.writelines to indicate which agent are working and which plugin that specific agents has used. However when i run with gpt-5 no agents are activated and no plugins are used.

To Reproduce
Steps to reproduce the behavior:

1. Create a simple magentic manager and add two or more agents to the magentic orchestration.
2. Invoke a prompt to the orchestration
3. See if the orchestration stalls and results in a timeout (expected)

Expected behavior
I expected the Magentic manager to select one or more agents to execute the tasks and result in response to the user.

Platform

• Language: C#
• Source: Semantic kernel ver. 1.67.1
• AI model: Azure Foundry OpenAI gpt-5
• IDE: JetBrains Rider
• OS: Windows 11

Additional context
Add any other context about the problem here.