StreamReason-Bench: Can Large Language Models Reason about Event-Time Stream-Processing Semantics?

StreamReason-Bench evaluates LLM accuracy in simulating event-time stream processing semantics using tumbling, hopping, and session windows. On a 600-item dataset, no model achieved over 34% exact match on event-time tasks, though chain-of-thought prompting roughly doubled performance for several models including GPT-4o.