[impossiblefork]

I don't think it's a common thing in any public LLM benchmarks or in any standard QA datasets. Maybe in internal stuff at AI firms.