[GH]score: 0.76vllm: Gemma4: route MoE with topk_softmax on XPUNovember 9, 2026The vLLM inference runtime now supports routing Mixture-of-Experts (MoE) models using topk_softmax on XPU hardware.read original ↗github.comshare this on XOne AI story a day, in plain English. The number that matters and what it means for you.FOLLOW @hackobar_DAILY DIGEST_all newsbuilderresearcherfounderinvestordesignerpolicyhealthsubscribe →you don't check 9 sources. we do. one email every morning, read in 2 min. free. unsubscribe anytime. privacy← back to feed