Give your team a private AI assistant with OpenWebUI and Ollama, running from a pre-configured AMI in your own AWS account. Chat with local or remote models while your prompts and documents stay on your infrastructure.
OpenWebUI is a team-friendly chat interface for large language models, and Ollama runs open models such as Llama, Mistral and Gemma locally. Together they give you a ChatGPT-style assistant you host yourself, with remote providers available too.
A clean UI for talking to local and remote LLMs.
Run Llama, Mistral, Gemma and GGUF or GPTQ variants with Ollama.
Built-in RAG over your own documents.
User accounts with granular permissions, managed by an admin.
Extend chats with functions and pipelines.
Keep conversations inside your own AWS account.
The AMI ships OpenWebUI and Ollama wired together, ready for the first admin account.
Subscribe on AWS Marketplace and launch the CloudFormation stack with your domain and admin email.
After 5-10 minutes, create a DNS record with the PublicIp from the stack outputs.
Open the DashboardUrl and create the admin account with a strong password.
Pull or connect a model, then add team members from the admin panel.
An AMI on AWS Marketplace, launched into your own AWS account with a CloudFormation stack.
Launch on AWS | OpenWebUI | LibreChat | |
|---|---|---|
| Licence | Open WebUI License (since April 2025) | MIT |
| Branding requirement | Yes, at 50+ users | None |
| Ollama integration | Deep, native | Through the OpenAI-compatible API |
| Built-in RAG | Yes | Through configuration |
| Providers out of the box | Ollama and OpenAI-compatible | OpenAI, Anthropic, Gemini, Bedrock and more |
OpenWebUI is source-available rather than OSI-approved open source. Read OpenWebUI vs LibreChat →
Local open models through Ollama, such as Llama, Mistral and Gemma, and remote models through compatible providers.
It is source-available. Since April 2025 it uses the Open WebUI License, which requires keeping its branding once a deployment reaches 50 or more users.
Open the DashboardUrl from the stack outputs and fill in the form. Later accounts are created from the admin panel.
Yes. It works with standard AWS VPC setups, including isolated environments.
t3.medium or larger. Larger local models need more memory and CPU.
Conversations and documents stay on your instance. Anything you send to a remote model provider goes to that provider.
Back up your server data, remove the old stack and launch the new version from AWS Marketplace.
Features, providers and the licensing change most reviews miss.
Pick a model to run behind your chat interface.
A private AI assistant in minutes.
We build and run self-hosted AI infrastructure every day, from model selection to RAG. Tell us what you need.
Contact us