https://docs.vllm.ai/en/stable/api/vllm/v1/attention/backends/mla/flashinfer_mla_sparse_sm120/
flashinfer_mla_sparse_sm120 - vLLM
flashinfermlasparsevllm
https://forums.developer.nvidia.com/t/new-bleeding-edge-vllm-docker-image-avarok-vllm-nvfp4-gb10-sm120/354231
New bleeding-edge vLLM Docker Image: avarok/vllm-nvfp4-gb10-sm120 - DGX Spark / GB10 Projects -...
Dec 11, 2025 - Running NVFP4 MoE Models Copy and paste this code to get a snappy SOTA qwen3-next model running on your DGX Spark at an NVFP4 quant: # Pull the pre-built...