Duaer

AI recipes

Run local models with Ollama in Duaer

A self-hosted Duaer can reach Ollama on the same machine or network and plug open models such as Llama and Qwen into AI Agents and RAG, keeping data on your servers.

Connect Ollama in Duaer

  1. Install Ollama on the server and pull a model, such as ollama pull qwen2.5 or ollama pull llama3.2.
  2. Create an Ollama credential in Duaer and set Base URL by deployment: http://localhost:11434 when Duaer runs directly on the same machine; http://host.docker.internal:11434 when Duaer runs in Docker; http://ollama:11434 in the same docker compose. Refer to Ollama credentials in Duaer.
  3. On Chat Model of an AI Agent or Basic LLM Chain, attach Ollama Chat Model and pick the pulled model from Model.
  4. Select Execute step to confirm it answers.

Fully local RAG

To keep vectors local too, use Embeddings Ollama instead of Embeddings OpenAI after ollama pull nomic-embed-text. The rest matches the Google Drive chatbot.

When it cannot connect

  • Ollama listens on 127.0.0.1 by default. When Duaer runs on another host or container, set OLLAMA_HOST=0.0.0.0 before starting Ollama.
  • When an AI Agent calls tools, pick a model with tool calling, such as qwen2.5 or llama3.1 and later.
  • Duaer cloud cannot reach localhost on your computer. To use it from the cloud, put Ollama behind an HTTPS reverse proxy and set API Key in the credential.

Questions

Can Duaer cloud use Ollama on my computer?

No. Duaer cloud cannot reach your localhost. Self-host Duaer, or expose Ollama over HTTPS with an API key.

Does Ollama in Duaer use AI credits?

No Duaer AI credits. Ollama runs on your machine; Duaer only calls it.

In this section