This Open WebUI app makes building with AI more accessible and productive by allowing you to easily interact with Ollama and OpenAI APIs.
Requires application load balancer
Running Open WebUI on FlexStack requires an Application Load Balancer which incurs an additional fee of around $16/month per environment. The ALB is necessary for communicating over websockets, since this is not supported by API Gateway HTTP APIs
Getting Started
There are 2 optional environment variables to add when deploying. OPENAI_API_KEY and OLLAMA_BASE_URL. Alternatively, both can be configured in the admin panel after deploying.
After deploying create an admin user that will be used to create additional users and groups.
By self-hosting this popular and robust interface for LLMs you gain the flexibility to utilize both proprietary and open source models. In addition, Open WebUI enables advanced AI workflows and share with team with team members.
For a full list of features and capabilities visit the Open WebUI docs.
Recommendations
Open WebUI is the interface layer that consumes LLM apis. Because GPU instances are generally very expensive, the easiest and most cost effective approach is to consume a pay per token API. DeepInfra and OpenRouter are good options.
For more advanced use cases or to experiment with Ollama locally, use a tunnel such as ngrok. Make sure to set the appropriate environment variable for Ollama as described in the Ollama docs.