Claude 3.7 is rumored to be Claude's upcoming new model that can switch between deep thinking like o1 and instant responses like a conventional LLM with the same set of weights. Resolves yes if Claude 3.7 exceeds Claude 3.5 new on LiveBench using its instant response setting and no otherwise.

Resolves N/A if the model turns out to not have an instant response mode at all