LibreChat: your own ChatGPT, your keys, no per-seat billing
An open source chat interface that connects several models at once. For teams it turns the cost from per user into per actual use, which is usually far less.
Short answer
LibreChat is an open source chat interface that you install on your own server and connect to several AI providers with your own API keys. Instead of paying a subscription for every person on the team, you pay for actual token usage. On teams where not everyone uses AI daily, the cost difference is large. It does require you to administer the server and keep an eye on what the keys are spending.
- Last updated
- April 20, 2026
- Last verified
- April 20, 2026
Antes de empezar necesitas
- A VPS with Docker and at least 2 GB of RAM
- At least one API key from a model provider
- A domain with HTTPS if more than one person will use it
The cost of AI subscriptions in a team is calculated per person, not per use. In a team of ten where three use it daily and seven use it now and then, you pay for ten.
LibreChat changes that math: a single installation, everyone inside it, and spending measured in tokens actually consumed.
1. Download and install
mkdir -p /opt/librechat && cd /opt/librechat
git clone https://github.com/danny-avila/LibreChat.git .
cp .env.example .env
2. Configure the keys
In the .env file, add the keys for the providers you are going to use and generate the session secrets:
OPENAI_API_KEY=your_key
ANTHROPIC_API_KEY=your_key
CREDS_KEY=random_string_of_32_bytes
JWT_SECRET=another_random_string
3. Bring the installation up
docker compose up -d
Go in through port 3080 and create the first account.
4. Close registration
By default anyone who reaches the URL can register. As soon as you have created your team's accounts, disable open registration in the .env:
ALLOW_REGISTRATION=false
It is the step most often forgotten, and the one that turns your installation into free AI for strangers, paid with your card.
What you lose compared to the official apps
You get no voice mode, no desktop apps, and none of the satellite features each provider adds to its own product. You get chat, history, several models and control of your data. For an internal team that is usually enough; for heavy personal use, you will probably prefer the official app.
Frequently asked questions
Is it cheaper than paying for subscriptions?
It depends on usage. Someone working with AI all day usually spends more on tokens than a subscription costs. A team where half the people use it occasionally spends much less. Do the math with your real consumption from the first month before cancelling anything.
Can I use several models at once?
Yes, and that is its biggest advantage: you switch models inside the same conversation. It is useful for comparing answers without repeating the context.
Does it store the conversations?
Yes, in your own MongoDB database. That is exactly what teams are after when they do not want their history living on third party servers.
How do I avoid a surprise bill?
Set spending limits in each API provider's dashboard. That is the first thing to do, before giving anyone access.
