Legal AI has entered a phase where the announcement of a benchmark result is indistinguishable from a product launch. DescrybeLM's perfect multistate bar exam score, Genie AI's Eidetic Intelligence claim to global benchmark leadership, Ivo's blind study advantage over Claude for Word — each arrived with methodology published and rubric disclosed, which is more than the industry's biggest players have offered. Harvey led a benchmarking study against human lawyers without opening its methodology to the same scrutiny it invited competitors to face. Thomson Reuters updated CoCounsel with capabilities it…