vLLM: DeepSeek V4 Pro serving can be 106x cheaper than Opus 5 on AgentX workloads
woosuk_k · x · 2026-09-09
- vLLM published a new blog, "vLLM x AgentX: Optimizing for Real-World Agentic Serving," walking through full-stack optimizations — architecture, framework and runtime — measured on SemiAnalysis' public AgentX agentic benchmark.
- Headline result: on the same AgentX workload, DeepSeek V4 Pro on vLLM can be up to 106x cheaper than Opus 5. All optimizations are open-sourced in vLLM.
- The blog candidly covers what worked and what didn't; the inferact team contributed and invites developers to try it on their own agents.
More from Infra
- Revenue per GPU can rise as token prices fall, argues Gavin Baker — GavinSBaker · 2026-10-27
- Anthropic S-1: only $19B total compute spend since inception, overbuild warning — bookwormengr · 2026-10-27
- CoreWeave, You.com and NVIDIA lift BrowseComp 22.9% in 3-week RL post-training sprint — RichardSocher · 2026-10-27
- Grid queues make power architecture an AI-factory decision; Exowatt and Redwood offer fixes — ingliguori · 2026-10-27
- Two CVSS 9.9 Wasmtime flaws allow sandbox escape, one ARM64-only since 2020 — basedjensen · 2026-10-27
- AI Infra Meme: While Countries Need TW-Scale Power for Datacenters, People Argue About Lightbulbs — mgostIH · 2026-10-27