@adam-
Who here has put the latest Qwen models through the ringer?
Considering running my nanoclaw instance with it.
I understand that it depends on the task, but I don't think every thing I do with OpenClaw needs to have the most expensive model solving for it.