Configure Provider and Model
Before creating bots, you need to configure at least one LLM provider and add chat and embedding models. This guide walks you through the Web UI.
Prerequisites
- Memoh is running (see Docker installation)
- You have logged in to the Web UI at http://localhost:8082
Step 1: Open the Models Page
Click Models in the left sidebar to open the Provider and Model configuration page.

The page has two panels:
- Left: Provider list and search
- Right: Selected provider's details and models (or an empty state if none selected)
Step 2: Add a Provider
Click the Add Provider button (with a plus icon) at the bottom of the left panel.

In the dialog, fill in:
| Field | Description |
|---|---|
| Name | A display name for this provider (e.g. my-openai, ollama-local) |
| API Key | Your API key. For local services like Ollama, you can use a placeholder like ollama |
| Base URL | The API base URL (e.g. https://api.openai.com/v1, http://localhost:11434/v1 for Ollama) |

Example — OpenAI:
- Name:
openai - API Key:
sk-... - Base URL:
https://api.openai.com/v1
Example — Ollama (local):
- Name:
ollama - API Key:
ollama - Base URL:
http://localhost:11434/v1
Click Add to save. The new provider appears in the left sidebar.
Step 3: Add Models
Select a provider from the left panel. The right panel shows the provider form and the model list.

Click Add Model to open the model creation dialog.
Fill in:
| Field | Description |
|---|---|
| Client Type | API protocol: openai-responses, openai-completions, anthropic-messages, or google-generative-ai |
| Type | chat or embedding |
| Model | Model ID (e.g. gpt-4, llama3.2, text-embedding-3-small) |
| Display Name | Optional display name |
| Dimensions | Required for embedding models (e.g. 1536 for OpenAI embeddings) |
| Multimodal | For chat models only — enable if the model supports images |
You need at least:
- One chat model (for conversation)
- One embedding model (for memory)
Add them under the same or different providers. For example:
- Chat:
gpt-4with client typeopenai-responses(OpenAI) orllama3.2with client typeopenai-completions(Ollama) - Embedding:
text-embedding-3-smallwith client typeopenai-completions(OpenAI) ornomic-embed-textwith client typeopenai-completions(Ollama)
Step 4: Edit or Delete
- Provider: After selecting a provider, you can edit Name, API Key, and Base URL in the right panel, or delete the provider.
- Model: Use the edit or delete actions on each model card in the list.
Next Steps
Once you have at least one chat model and one embedding model, you can create a bot (via Bots in the sidebar) and assign these models in the bot settings.
