Auditing Claude Usage with the Copilot Researcher Agent

The question was asked if it was possible to identify use of the Claude LLM by the Copilot Researcher Agent. Audit records often help, so that's the natural location to check. As it turns out, some information is captured when the Researcher agent is used, but figuring out if the agent uses the default ChatGPT-5 or Claude LLMs is a matter of intuition (or guesswork).