[blooalien]

I have had nothing but good results using the Qwen2.5 and Hermes3 models. The response times and token generation speeds have been pretty fantastic compared against other models I've tried, too.