Hi! I think many LLM model experimenters have a 5090 and at least 128GB of host RAM; wouldn't it be a significant turning point to try and get an nvidia/Qwen3.5-122B-A10B-NVFP4 working? ... I suppose Qwen3.8-Flash-Next-NVFP4 is functionally impossible.
Hi! I think many LLM model experimenters have a 5090 and at least 128GB of host RAM; wouldn't it be a significant turning point to try and get an nvidia/Qwen3.5-122B-A10B-NVFP4 working? ... I suppose Qwen3.8-Flash-Next-NVFP4 is functionally impossible.