谷歌TPU跑Kimi比英伟达GPU快57%，底层用的是DeepSeek推理框架
谷歌TPU跑Kimi比英伟达GPU快57%！用的还是DeepSeek推理框架
vLLM那帮人创业后干的活：把Kimi K2塞进谷歌TPU，速度比英伟达GPU还快57%，框架还是DeepSeek开源的，搞推理的可以看看细节。
vLLM团队成员创立的创业公司Noop Labs展示了在谷歌TPU上运行Kimi K2的效果，吞吐量比英伟达GPU高出57%。
推理框架选用了DeepSeek开源的vLLM方案，适配TPU后无需改动模型代码。这意味着国产开源模型可以在谷歌自研芯片上高效运行，推理成本路径多了一个选择。
谷歌TPU跑Kimi比英伟达GPU快57%！用的还是DeepSeek推理框架
vLLM人马创业公司团队出品
This report is categorized under AITOP's "Models" desk, focusing on China's AI industry and domestic foundation model ecosystem. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows and US-China foundation model benchmarks and commercialization comparison. Key entities and technical topics under continuous tracking include Kimi K2, Google TPU, vLLM, DeepSeek.
Source attribution: curated from 量子位 (@听雨).
