Home / Servers / LibreChat VPS

Self-host LibreChat on your own VPS

LibreChat is the second most-starred self-hosted ChatGPT alternative, with deeper provider support than any competitor: OpenAI, Anthropic, AWS Bedrock, Azure, Google Vertex, Groq, Mistral, OpenRouter, plus any OpenAI-compatible endpoint including Ollama. Hosting it on your own VPS keeps prompts off third-party servers and gives you a single private interface for every LLM you use. Servury is built for it: anonymous signup, crypto payments, full root, no logs.

LibreChat needs at least 4 GB RAM. Starting at $15.59/mo.

Why self-host LibreChat

Most providers in one UI. Switch between Claude, GPT-4, Gemini, Llama, Mistral, and dozens more in the same conversation, or per-conversation.
Your prompts stay local. Conversations and uploaded files live in MongoDB on your VPS. No third party except the LLM endpoint sees your data.
Multi-user, group-ready. Roles, group permissions, conversation sharing. Run a private team chat without the SaaS bill.
Anonymous infra. Pay with XMR or BTC, no email required. Your private LLM hub stays unlinked from your real identity.

Quick start: LibreChat on Servury via Docker

Tested on Ubuntu 24. Pick a 4 GB+ plan, deploy, SSH in.

# 1. Install Docker + Compose plugin
curl -fsSL https://get.docker.com | sh
apt install -y docker-compose-plugin git

# 2. Clone LibreChat and start the stack
git clone https://github.com/danny-avila/LibreChat.git
cd LibreChat
cp .env.example .env

# 3. Edit .env to add your API keys (OpenAI, Anthropic, etc.)
#    Set HOST=0.0.0.0 to bind to all interfaces.
nano .env

# 4. Bring up the stack (LibreChat + MongoDB + Meilisearch)
docker compose up -d

# 5. Open http://YOUR_SERVER_IP:3080 to register the first admin
# 6. Recommended: put Caddy or nginx in front for HTTPS

Recommended specs

Solo / personal

Just you, cloud LLM backends, light usage.

4 GB RAM · 2 vCPU · 40 GB SSD
VPS-150 plan
Recommended

Small team, multi-provider, growing chat history and search index.

8 GB RAM · 4 vCPU · 80 GB SSD
VPS-200 plan
Production

Many users, large vector index, file uploads at scale.

16+ GB RAM · 8 vCPU · 160 GB SSD
VPS-250 or VDS

Frequently asked questions

What is LibreChat?

LibreChat is an open-source self-hosted ChatGPT alternative that supports more LLM providers than any other chat interface: OpenAI, Anthropic, AWS Bedrock, Azure, Google Vertex AI, Groq, Mistral, OpenRouter, and any OpenAI-compatible endpoint including Ollama. It is the second most-starred self-hosted chat UI after Open WebUI.

How is LibreChat different from Open WebUI?

Both are excellent. LibreChat has wider built-in provider support (especially enterprise ones like Bedrock and Vertex), more advanced multi-user/permission tooling, and richer plugin support. Open WebUI has a simpler UI and cleaner Ollama integration. Pick based on which providers you actually use.

How much VPS do I need?

LibreChat ships several services in one Docker Compose stack (LibreChat API, MongoDB, Meilisearch). 4 GB RAM is the practical minimum, 8 GB is the sweet spot for any team use.

Can I use my own API keys?

Yes. LibreChat is designed to bring-your-own-keys. You set them in the .env file (or per-user via the admin panel) and they live on your VPS, not in any cloud.

Does it support file uploads and document chat?

Yes. Drop in PDFs, images, code files. LibreChat handles ingestion and gives you a chat-with-document experience. Files are stored locally on your VPS.

Can I connect a local Ollama instance?

Yes. Add Ollama as a custom OpenAI-compatible endpoint in the config. Many users run LibreChat + Ollama on the same VPS for fully-local chat.

Will my data stay private?

Yes if you self-host. All conversations, uploads, and search indexes are stored locally on your VPS. Servury runs zero application-level logging on customer servers.

How do I update LibreChat?

cd LibreChat, git pull, docker compose pull, docker compose up -d. Mongo data persists across upgrades. Always read the release notes for breaking config changes between major versions.

Where are servers located?

Montreal (owned hardware), New York, London, Paris, Frankfurt, Netherlands, and Singapore. Pick whichever is closest to your users.