I’ve had awesome results with Qwen3-30B-A3B compared to other local LMs I’ve tried. Still not crazy good but a lot better and very fast. I have 24GB of VRAM though