own/metal

Self-host the AI stack
you actually control.

Find self-hosted alternatives to popular AI tools. Each tool comes with verified RAM, CPU and GPU requirements, plus the exact VPS size you need to run it.

[ FEATURED ]

Where we'd start.

Battle-tested. Well-documented. Kindly licensed. Specs verified on a real deployment.

[ CATEGORIES ]

By stack layer.

Every tool grouped by what it does in your stack — from the runtime to the UI on top.

[ INDEX ]

All entries.

Chat UIs CPU

AnythingLLM

Chat with your documents — workspaces, embeddings, agents in one container.

RAM 4 GB
vCPU 2
MIT setup · 1/5
Image Generation GPU

AUTOMATIC1111 WebUI

The classic Stable Diffusion web UI.

VRAM 8 GB
RAM 16 GB
AGPL-3.0 setup · 3/5
Vector Databases CPU

Chroma

The simplest vector database for prototyping.

RAM 1 GB
vCPU 1
Apache-2.0 setup · 1/5
Image Generation GPU

ComfyUI

Node-based UI for Stable Diffusion, Flux and beyond.

VRAM 12 GB
RAM 16 GB
GPL-3.0 setup · 3/5
Coding Assistants CPU

Continue

Open-source Copilot for VS Code and JetBrains — bring your own model.

RAM 1 GB
vCPU 1
Apache-2.0 setup · 1/5
Agents & Workflows CPU

Dify

Self-hosted LLM app platform — datasets, agents, observability.

RAM 8 GB
vCPU 4
Dify Open Source setup · 3/5
Agents & Workflows CPU

Flowise

Drag-and-drop builder for LangChain flows.

RAM 2 GB
vCPU 2
Apache-2.0 setup · 1/5
Chat UIs CPU

Hollama

Minimal, fast Ollama and OpenAI client. No backend.

RAM 512 MB
vCPU 1
MIT setup · 1/5
Image Generation GPU

InvokeAI

Polished, production-leaning Stable Diffusion studio.

VRAM 8 GB
RAM 16 GB
Apache-2.0 setup · 2/5
RAG & Knowledge CPU

Khoj

Personal AI that searches your notes, files and the web.

RAM 4 GB
vCPU 2
AGPL-3.0 setup · 2/5
Agents & Workflows CPU

Langflow

Visual flow builder for LangChain — IBM-backed.

RAM 4 GB
vCPU 2
MIT setup · 2/5
Chat UIs CPU

LibreChat

Multi-provider chat UI with agent and tool support.

RAM 4 GB
vCPU 2
MIT setup · 2/5
LLM Runners CPU

llama.cpp

The C/C++ LLM inference engine that runs everywhere.

RAM 8 GB
vCPU 4
MIT setup · 3/5
LLM Runners CPU

LocalAI

Drop-in OpenAI-compatible API for local models.

RAM 16 GB
vCPU 6
MIT setup · 2/5
Agents & Workflows CPU

n8n

Workflow automation with first-class AI nodes.

RAM 2 GB
vCPU 2
Sustainable Use setup · 1/5
LLM Runners CPU

Ollama

The simplest way to run open-weight LLMs locally.

RAM 8 GB
vCPU 4
MIT setup · 1/5
RAG & Knowledge CPU

Onyx

Self-hosted enterprise search and chat over your team's data.

RAM 16 GB
vCPU 8
MIT setup · 4/5
Chat UIs CPU

Open WebUI

The leading self-hosted ChatGPT alternative.

RAM 2 GB
vCPU 2
MIT setup · 1/5
AI Search CPU

Perplexica

Self-hosted Perplexity alternative — AI search with citations.

RAM 2 GB
vCPU 2
MIT setup · 2/5
Voice & Audio CPU

Piper

Fast, lightweight neural text-to-speech.

RAM 512 MB
vCPU 1
MIT setup · 1/5
Vector Databases CPU

Qdrant

Fast Rust-based vector database with great DX.

RAM 2 GB
vCPU 2
Apache-2.0 setup · 1/5
Coding Assistants GPU

Tabby

Self-hosted GitHub Copilot — full backend included.

VRAM 8 GB
RAM 8 GB
Apache-2.0 setup · 2/5
LLM Runners GPU

vLLM

High-throughput LLM serving for production GPU workloads.

VRAM 24 GB
RAM 32 GB
Apache-2.0 setup · 3/5
Vector Databases CPU

Weaviate

Vector DB with built-in hybrid search and modular embeddings.

RAM 4 GB
vCPU 2
BSD-3-Clause setup · 2/5
Voice & Audio CPU

whisper.cpp

OpenAI Whisper in C/C++ — CPU-friendly transcription.

RAM 4 GB
vCPU 4
MIT setup · 2/5