Analogical reasoning is one of the clearest tests of whether an LLM is truly reasoning or merely performing sophisticated pattern matching. Analogies require identifying structural relationships that remain constant even when surface details change. Evaluating models with analogical tasks is a stress test for cognitive depth: whether a model generalizes beyond wording and superficial cues, or breaks when those cues shift. A model that fails should not be trusted with legal reasoning or scientific discovery, where recognizing deep structural parallels across cases is essential.