ai-agent

paddler: Scale LLMs

Founders get a simple load balancer for self-hosting large language models with paddler, used by the community with 1.6k+ GitHub stars

From the repo ยท โ€œOpen-source LLM/VLM load balancer and serving platform for self-hosting LLMs (and VLMs) at scale ๐Ÿ“๐Ÿฆ™ Alternative to projects like llm-d, Docker Model Runner, etc but with less moving parts and simple deployments built around ggml ecosystem. Runs on CPU and GPU.โ€

intermediateโฑ 1-2 hours๐Ÿ’ต Free (self-hosted)
1,681 stars+6โ˜… this week97 forksRustGuide quality 9/10 ยท reviewed Jun 30Updated 10/21/2026100% free ยท open source
What it is

You can self-host Large Language Models (LLMs) with paddler to create highly customized applications.

Real use case example

"A solo developer is creating a chatbot for their startup's customer support. They use paddler to host their LLM, set up load balancing, and serve queries from their users. They configure paddler to pull product data from Amazon to answer customers' product-related questions. With paddler, their chatbot can now provide more accurate and helpful responses."

Best insideSelf-hosted
When to use it
  • โ€ขWhen you need to deploy and manage multiple LLMs or VLMs in a self-hosted environment
  • โ€ขWhen you require a simple and lightweight load balancer for your AI models
  • โ€ขWhen you want to use a ggml ecosystem-based solution for deploying and serving AI models
Ready-to-paste prompt
curl -X POST 'http://localhost:8080/infer' -H 'Content-Type: application/json' -d '{"model": "my_llm", "prompt": "Write a story about a character who learns a new skill"}'
Heads up: Make sure you have Rust and Cargo installed on your system, as paddler is built using Rust and requires these tools to build and run
Saves to your device
Use with Claude
New

Skip the builder โ€” one click puts this in Claude, Cursor, Antigravity and more.

โœ… Light setup

Installs with a command or two; your AI agent can do it for you.

Try it instantly โ€” no install
Claude Code
mkdir -p ~/.claude/skills/paddler && curl -fsSL https://workflowstacks.com/api/skills/paddler/claude-skill -o ~/.claude/skills/paddler/SKILL.md
Open in another AI app

Opens the app with this repo with the prompt ready to go โ€” no copy-paste needed.

Connect the whole catalog (MCP)
claude mcp add --transport http workflowstacks https://workflowstacks.com/api/mcp

Adds a WorkflowStacks connector to Claude Code: search and load any skill here by chatting.

How paddler: Scale LLMs works
Codeflow
Free to inspect

Paddler: Scale LLMs is a large Rust project (~46k lines across 902 code files, plus 487 test files). Setup is light: runs with Docker. Last commit this month, Apache-2.0 license, has a test suite.

Size
Large codebase
~46k lines ยท 902 code files ยท ~6 h to skim
Setup
Light setup
Runs with Docker.
Runs on
Rust ยท Docker
No API keys detected
Rust 81%TypeScript 9%Python 7%CSS 1%Nix 1%
Where to start reading
  1. 1
    README.md
    Start here โ€” what it does and how to install it
  2. 2
    CLAUDE.md
    The instructions the AI actually follows
  3. 3
    Cargo.toml
    Dependencies and the commands it exposes
  4. 4
    example/README.md
    A worked example โ€” copy this to get going
What's in each folder
example/Examples you can copy2 files
paddler_tests/Folder275 files
paddler_balancer/Folder189 files
resources/Folder140 files
paddler_client_python/Folder127 files
paddler_agent/Folder105 files
paddler_messaging/Folder95 files
READMEHas testsDocumentedCI checksDocker readyExamples includedApache-2.0 licenseUpdated this month
Quick Actions
Details
Creator
intentee
Language
Rust
Category
ai-agent
Published
4/27/2024

Are you the creator of this tool? and earn 85% of every sale.

Show it off in your README:

Featured on WorkflowStacks
[![Featured on WorkflowStacks](https://workflowstacks.com/api/badge/paddler.svg)](https://workflowstacks.com/skills/paddler?utm_source=github&utm_medium=badge)
๐Ÿ”ฅ Hot this week, in your inbox
Skills like this, every Monday.

The five fastest-growing open-source AI skills, ranked by GitHub star growth. One email a week, unsubscribe anytime.

Do this with a working workflow

Pages we wrote and tested that put a skill like this to work โ€” importable n8n templates, outcome guides and Claude Desktop configs.