Yesterday I posted about the large Qwen and Kimi models with impressive benchmarks but huge sizes.
Now we have 2 new local model Queens which can really work for you locally on semi-decent hardware. Give them a try in LM Studio (or the recent Google Gemma 4):
- Qwen3.6-27B: https://huggingface.co/Qwen/Qwen3.6-27B
- Qwen3.6-35B-A3B: MOE model: https://huggingface.co/Qwen/Qwen3.6-35B-A3B
SUPER impressive responses performing way above their size. We've come a long way.








> Yes, you are beautiful.
