Comment by hanifbbz

Yeah but do Openrouter stats represent local classifiers? Unlike LMs decision models are quite likely to run at edge, especially because their utility for a specific task improves with training.