vLLM matches DeepSeek's DeepInfer decode performance within 6%, generating $39/GPU/hr at off-peak pricing
vllm_project · x · 2026-10-21
The vLLM team reports that with Wide-EP and EPLB, vLLM can reproduce DeepSeek's published DeepInfer decode throughput: under DeepSeek's decode configuration (DeepSeek-V4.1-Flash, 131K context, nextn=3), vLLM on GB200 stays within 6% of DeepSeek's numbers, translating to $39/GPU/hr in revenue at DeepSeek's off-peak pricing (AgentX workload, 200 P90 TPS).
The key: vLLM has continuously integrated DeepSeek's kernel libraries (MegaAttn, MegaMoE, Mega-mHC, etc.) and optimized the rest of the stack, so with largely identical kernels the two engines perform about the same. Detailed numbers and reproduction recipes are public.
More from Infra
- vLLM's specialized decode path hits 829 P90 TPS per user on four B300 GPUs, 26% faster — vllm_project · 2026-10-21
- vLLM says it can reproduce DeepSeek's inference performance, yielding $39/GPU/hr revenue — vllm_project · 2026-10-21
- AMD brings EAGLE-4 speculative decoding to Instinct MI355X day 0, 1.6x faster than built-in MTP — hongyangzh · 2026-10-21
- TokenGate: learned token pruning speeds up LLM prefill up to 11.3x — Liana Patel · 2026-10-21
- Modal revives multiplayer game Curvytron to demo its new Sticky Sessions routing — charles_irl · 2026-10-21
- Seroter's reading list: TorchTPU open source, Claude Code's code review system — rseroter · 2026-10-21