AI Open Chat Playground

This provides a web UI for AI chat playground that is able to connect virtually any LLM from any platform.

Prerequisites

• .NET SDK 9
• Visual Studio Code + C# DevKit
• Docker Desktop or Podman
• PowerShell 7.4+
• Azure Developer CLI
• Azure CLI + Container Apps extension
• GitHub CLI

Getting Started

Get the repository ready

1. Login to GitHub.

   gh auth login

2. Check login status.

   gh auth status

3. Fork this repository to your account and clone the forked repository to your local machine.

   gh repo fork aliencube/open-chat-playground --clone --default-branch-only

Run it locally

1. Get the repository root.

   # bash/zsh
   REPOSITORY_ROOT=$(git rev-parse --show-toplevel)

   # PowerShell
   $REPOSITORY_ROOT = git rev-parse --show-toplevel

2. Run the app.

   dotnet run --project $REPOSITORY_ROOT/src/OpenChat.AppHost

3. If you want to change the model, pass the following arguments:

   dotnet run --project $REPOSITORY_ROOT/src/OpenChat.AppHost -- [OPTIONS]

   Here are the list of options:

   • --llm-provider: Choose openai, ollama or hface. Default is openai.
   • --openai-deployment: Provide the deployment name, if you choose openai as the LLM provider. Default is gpt-4o.
   • --openai-connection-string: Provide the connection string to OpenAI. It must be provided if you choose openai as the LLM provider. It must follow the format like Endpoint=xxxxx;Key=xxxxx.
   • --ollama-image-tag: Provide the Ollama container version, if you choose either ollama or hface as the LLM provider. Default is 0.6.8.
   • --ollama-use-gpu: Provide the value whether to use the GPU acceleration or not, if you choose either ollama or hface as the LLM provider. Default is false.
   • --ollama-deployment: Provide the deployment name, if you choose ollama as the LLM provider. Default is llama.
   • --ollama-model: Provide the model name, if you choose ollama as the LLM provider. Default is llama3.2.
   • --huggingface-deployment: Provide the deployment name, if you choose hface as the LLM provider. Default is qwen3.
   • --huggingface-model: Provide the model name, if you choose hface as the LLM provider. Default is Qwen/Qwen3-14B-GGUF.
   • --help: Display the help message.

Run on Azure

1. Get the repository root.

   # bash/zsh
   REPOSITORY_ROOT=$(git rev-parse --show-toplevel)

   # PowerShell
   $REPOSITORY_ROOT = git rev-parse --show-toplevel

2. Make sure you are at the repository root.

   cd $REPOSITORY_ROOT

3. Login to Azure.

   # Login to Azure Dev CLI
   azd auth login
   
   # Login to Azure CLI
   az login

4. Check login status.

   # Azure Dev CLI
   azd auth login --check-status
   
   # Azure CLI
   az account show

5. Update appsettings.json on the OpenChat.AppHost project. The JSON object below shows the default values.

   {
     "LLM": {
       // Set the LLM provider.
       "Provider": "openai"
     },
   
     "OpenAI": {
       // Set the deployment name, if you choose 'openai' as the LLM provider.
       "DeploymentName": "gpt-4o"
     },
   
     "Ollama": {
       // Set the Ollama container image tag.
       "ImageTag": "0.6.8",
       // Set to either 'true' or 'false' depending on the usage of the GPU acceleration.
       "UseGPU": false,
       // Set the deployment name, if you choose either `ollama` or `hface` as the LLM provider.
       "DeploymentName": "llama",
       // Set the model name, if you choose either `ollama` or `hface` as the LLM provider.
       // Make sure the model name must include '/' in the middle and 'GGUF' at the end, if you choose 'hface' as the LLM provider.
       "ModelName": "llama3.2"
     }
   }

6. Add OpenAI connection string, if you want to use OpenAI as the LLM provider.

   # bash/zsh
   dotnet user-secrets --project $REPOSITORY_ROOT/src/OpenChat.AppHost \
       set ConnectionStrings:openai "Endpoint={{API_ENDPOINT}};Key={{API_KEY}}"

   # PowerShell
   dotnet user-secrets --project $REPOSITORY_ROOT/src/OpenChat.AppHost `
       set ConnectionStrings:openai "Endpoint={{API_ENDPOINT}};Key={{API_KEY}}"

7. Run the following commands in order to provision and deploy the app.

   azd up

   NOTE: You will be asked to provide Azure subscription and location for deployment.