Glossary · Term

vLLM

← all terms

Definition

Plain language

A popular open-source system for serving large language models efficiently.

As stated in the literature

An open-source high-throughput inference engine for large language models, widely used as a default open-source serving stack.

Why it matters: It is the default high-throughput serving stack much of the open-source LLM ecosystem builds on, so its performance affects countless deployments.

For example, a startup might launch its chatbot on vLLM running on a few A100 GPUs to handle thousands of concurrent users.

Heard on the show

“Below that, a serving framework — vLLM and SGLang are the two big open-source ones — which batches users together and manages memory.”
Episode 139 — When Optimizing One GPU Kernel Quietly Breaks the Whole System

Mentioned in 3 episodes

  1. 139
    When Optimizing One GPU Kernel Quietly Breaks the Whole System
  2. 027
    When AI Agents Build the Serving Stack: A Bet on Bespoke Infrastructure
  3. 016
    Why Your Coding Agent Stalls While the GPU Runs Hot

Related terms