AI & LLM apps you can self-host — Oct 2026

Every ai & llm tool below runs on your own server — no per-seat bills, no vendor lock-in.

Rather than a flat list, we ranked these 16 tools by how many times people actually started them on PickYour.App.

How they stack up

#AppCategoryInstallsFrom
1Ollama
Run Llama, Mistral & Gemma models locally
AI & LLM94 000€0.092/hr
2Open WebUI
ChatGPT-style interface for your LLMs
AI & LLM82 000€0.092/hr
3OpenHands
Self-hosted AI developer control center — run coding agents across local, remote & cloud
AI & LLM78 000€0.046/hr
4ComfyUI
Node-based UI for diffusion image workflows
AI & LLM62 000€0.092/hr
5AnythingLLM
All-in-one AI app — chat, documents & agents
AI & LLM56 000€0.046/hr
6Dify
Production-ready LLM app platform
AI & LLM54 000€0.092/hr
7Flowise
Drag-and-drop LLM orchestration
AI & LLM49 000€0.046/hr
8Qdrant
High-performance vector database for AI
AI & LLM47 000€0.046/hr
9LibreChat
Open-source ChatGPT alternative for every LLM
AI & LLM42 000€0.046/hr
10bolt.diy
AI full-stack web builder in the browser — open-source Bolt.new
AI & LLM41 000€0.046/hr
11LiteLLM
One API to call 100+ LLM providers, OpenAI-compatible
AI & LLM39 000€0.046/hr
12Tabby
Self-hosted AI coding assistant — the open-source Copilot alternative
AI & LLM36 000€0.046/hr
13Perplexica
Open-source AI search engine, Perplexity alternative
AI & LLM34 000€0.046/hr
14LocalAI
Drop-in OpenAI API replacement that runs locally
AI & LLM31 000€0.092/hr
15Langfuse
LLM engineering: traces, evals & prompt management
AI & LLM28 000€0.046/hr
169router
Free AI-coding gateway for Claude Code, Codex, Cursor & more — 40+ providers with auto-fallback
AI & LLM21 000€0.023/hr

Reading the field

Ollama

Teams pick Ollama when they need the basics done well. Spec: 4 vCPU, 8 GB RAM, 100 GB disk. Start here if you want the safe pick.

Open WebUI

Open WebUI covers the essentials without demanding a manual. Spec: 4 vCPU, 8 GB RAM, 100 GB disk. Close second, and some teams prefer its take on the problem.

OpenHands

Built around its core job, OpenHands stays out of your way. It runs comfortably on 2 vCPU / 4 GB here. Newer to the catalog, climbing steadily.

ComfyUI

The core loop is what ComfyUI does best. Plan for 4 vCPU and 8 GB of RAM. Newer to the catalog, climbing steadily.

Before you install

None of these need a big machine. Budget roughly €0.092/hour for the largest spec in the table (peak 4 vCPU, 8 GB), and less for the rest. The install wizard pre-selects the right plan for you.

Generated 2026-10-22 from live catalog data · all guides