---
title: "Discussion on \"Building a distributed tokens-per-minute limiter for LLM APIs with one Redis Lua script\" | Hashnode"
description: "Discussion on \"Building a distributed tokens-per-minute limiter for LLM APIs with one Redis Lua script\". Token buckets, reserve-and-settle leases, crash-safe concurrency slots, and why a 429 isn't always a rate limit"
url: "https://hashnode.com/posts/building-a-distributed-tokens-per-minute-limiter-for-llm-apis-with-one-redis-lua-script/6af1c6a6f7522e005a9755d2"
---

[Feed](https://hashnode.com/feed) /

Discussion

# [Building a distributed tokens-per-minute limiter for LLM APIs with one Redis Lua script](https://quietkernel.hashnode.dev/building-a-distributed-tokens-per-minute-limiter-for-llm-apis-with-one-redis-lua-script?utm_source=hashnode&utm_medium=feed)

Token buckets, reserve-and-settle leases, crash-safe concurrency slots, and why a 429 isn't always a rate limit

1d ago · 7 min read · quietkernel.hashnode.dev  
Read the article

-   [#typescript](https://hashnode.com/tag/typescript)
-   [#nodejs](https://hashnode.com/tag/nodejs)
-   [#redis](https://hashnode.com/tag/redis)
-   [#llm](https://hashnode.com/tag/llm)
-   [#backend](https://hashnode.com/tag/backend)

## Written by

[Krrish Kumar](https://hashnode.com/@quietkernel)

@quietkernel

  
[View profile](https://hashnode.com/@quietkernel)

Krrish Kumar, backend & AI engineer (Node.js/TypeScript)

Krrish Kumar, backend & AI engineer (Node.js/TypeScript). I maintain secret-blast-radius, create-bullmq-app, zod-india and webhook-crash-lab.

## Responses

Markdown supported · ⌘↵ to post

No responses yet.