Comment by gliched_robot

Inference speed is not a great metric given the horizontal scalability of LLMs.