As far as I can tell, there are no benchmarks at all for Grok 4.2, is that right? Every other pre-release model I have seen had at least some benchmarks. Even if they are flawed (and benchmarks are flawed), they help you see the tasks for which the model might be worth trying.