@burkeholland Not sure if nvfp4 quants work properly yet so I’d suggest these unsloth GGUFs:
Big single model: unsloth/Qwen3.8-Flash-Next-GGUF Q4_K_M
Smaller model (subagent-OK): unsloth/Qwen3.8-27B-GGUF UD-Q4_K_M
@0xBakeer I’m really liking the new nvidia/Qwen3.8-Flash-Next-FP4-Local on my Spark. Are you running it on high concurrency or do you just prefer the 27B?
@0xBakeer Fax. These 128gb RTX laptops make no sense. Doesn’t have to be a cheap one tho, I prefer a nice MacBook Air (or similar) with good battery life.