Ollama Chat Client Example

Introduction

Simple .NET 9 chat client/api example that demonstrates how to use the Ollama Chat API.

Getting Started

• Set up persistent volume local folders and update docker-compose.yml with folder locations first.
• docker-compose up -d to pull and start Ollama and Redis docker containers.
• docker exec -it ollama ollama pull llama3 to pull the llama3 model into the Ollama container (this will take a while).
• If using VS Code RUN AND DEBUG > Launch Debug > Run
• Switching between emulator and Ollama AI Chat Client is done in appSettings.development.json

  "ChatSettings": {
    "UseEmulator": true,
    "EmulatorResponseTime": 10000 // milliseconds
  }

Things to Note

• I am using Redis to store chat history, and this has been 'abused' to manage the chat history.
• The ChatService should ideally be implemented as a scalable background microservice. I am using DotNet Channel for background process management because this is a monolithic example.
• Throttling/request scaling has not been implemented as I am using a non-distributed Chat API (Ollama in this case), so if you throw multiple requests at Ollama while it is still trying to figure things out, you will run into issues.
• Redis has been configured (in code) to invalidate conversations after 15 mins.
• Plenty more that can be done. This is not production code, but a simple example to demonstrate how to use the Ollama Chat API.
• ChatClientEmulator is a simple IChatClient replacement so we can bypass requests to Ollama. It is there to ease development.

Other Useful Information

• docker exec -it ollamma bash to enter the container.`
• You are not confined to llama3. You can pull other models as well, such as DeepSeek-R1, Phi-4, Mistral, Gemma 3, and more. Visit https://ollama.ai/models for more information.