dotnet · luisquintanilla · Jun 25, 2026 · Jul 8, 2026 · Jul 8, 2026 · Jul 8, 2026
@@ -19,6 +19,7 @@
     <PackageVersion Include="CommunityToolkit.Aspire.Hosting.Ollama" Version="13.0.0-beta.444" />
     <PackageVersion Include="CommunityToolkit.Aspire.OllamaSharp" Version="13.0.0-beta.444" />
     <PackageVersion Include="MessagePack" Version="2.5.301" />
+    <PackageVersion Include="Microsoft.AI.Foundry.Local" Version="1.2.3" />
     <PackageVersion Include="Microsoft.Agents.AI" Version="1.3.0" />
     <PackageVersion Include="Microsoft.Agents.AI.DevUI" Version="1.3.0-preview.260423.1" />
     <PackageVersion Include="Microsoft.Agents.AI.Hosting" Version="1.3.0-preview.260423.1" />

@@ -128,18 +128,13 @@
       "type": "parameter",
       "displayName": "_AI service provider",
       "datatype": "choice",
-      "defaultValue": "githubmodels",
+      "defaultValue": "foundrylocal",
       "choices": [
         {
           "choice": "azureopenai",
           "displayName": "Azure OpenAI",
           "description": "Uses Azure OpenAI service"
         },
-        {
-          "choice": "githubmodels",
-          "displayName": "GitHub Models",
-          "description": "Uses GitHub Models"
-        },
         {
           "choice": "ollama",
           "displayName": "Ollama (for local development)",
@@ -149,6 +144,11 @@
           "choice": "openai",
           "displayName": "OpenAI Platform",
           "description": "Uses the OpenAI Platform"
+        },
+        {
+          "choice": "foundrylocal",
+          "displayName": "Foundry Local (on-device)",
+          "description": "Uses Foundry Local with the qwen3-4b and qwen3-embedding-0.6b models"
         }
         // {
         //   "choice": "azureaifoundry",
@@ -215,14 +215,14 @@
       "type": "computed",
       "value": "(AiServiceProvider == \"openai\")"
     },
-    "IsGHModels": {
-      "type": "computed",
-      "value": "(AiServiceProvider == \"githubmodels\")"
-    },
     "IsOllama": {
       "type": "computed",
       "value": "(AiServiceProvider == \"ollama\")"
     },
+    "IsFoundryLocal": {
+      "type": "computed",
+      "value": "(AiServiceProvider == \"foundrylocal\")"
+    },
     "IsAzureAIFoundry": {
       "type": "computed",
       "value": "(AiServiceProvider == \"azureaifoundry\")"
@@ -317,6 +317,38 @@
       },
       "replaces": "all-minilm"
     },
+    "FoundryLocalChatModelDefault": {
+      "type": "generated",
+      "generator": "constant",
+      "parameters": {
+        "value": "qwen3-4b"
+      }
+    },
+    "FoundryLocalEmbeddingModelDefault": {
+      "type": "generated",
+      "generator": "constant",
+      "parameters": {
+        "value": "qwen3-embedding-0.6b"
+      }
+    },
+    "FoundryLocalChatModel": {
+      "type": "generated",
+      "generator": "coalesce",
+      "parameters": {
+        "sourceVariableName": "ChatModel",
+        "fallbackVariableName": "FoundryLocalChatModelDefault"
+      },
+      "replaces": "qwen3-4b"
+    },
+    "FoundryLocalEmbeddingModel": {
+      "type": "generated",
+      "generator": "coalesce",
+      "parameters": {
+        "sourceVariableName": "EmbeddingModel",
+        "fallbackVariableName": "FoundryLocalEmbeddingModelDefault"
+      },
+      "replaces": "qwen3-embedding-0.6b"
+    },
     "webHttpPort": {
       "type": "parameter",
       "datatype": "integer",

@@ -1,17 +1,11 @@
 var builder = DistributedApplication.CreateBuilder(args);
 #if (IsOllama) // ASPIRE PARAMETERS
-#elif (IsOpenAI || IsGHModels)
+#elif (IsOpenAI)
 
 // You will need to set the connection string to your own value
 // You can do this using Visual Studio's "Manage User Secrets" UI, or on the command line:
 //   cd this-project-directory
-#if (IsOpenAI)
 //   dotnet user-secrets set ConnectionStrings:openai "Key=YOUR-API-KEY"
-#elif (IsGHModels)
-//   dotnet user-secrets set ConnectionStrings:openai "Endpoint=https://models.inference.ai.azure.com;Key=YOUR-API-KEY"
-#else // IsAzureOpenAI
-//   dotnet user-secrets set ConnectionStrings:openai "Endpoint=https://YOUR-DEPLOYMENT-NAME.openai.azure.com;Key=YOUR-API-KEY"
-#endif
 var openai = builder.AddConnectionString("openai");
 #else // IsAzureOpenAI
 
@@ -62,7 +56,7 @@
     .WithReference(embeddings)
     .WaitFor(chat)
     .WaitFor(embeddings);
-#elif (IsOpenAI || IsGHModels)
+#elif (IsOpenAI)
 webApp.WithReference(openai);
 #else // IsAzureOpenAI
 webApp

@@ -7,19 +7,29 @@
     <UserSecretsId>d5681fae-b21b-4114-b781-48180f08c0c4</UserSecretsId>
   </PropertyGroup>
 
+<!--#if (IsFoundryLocal) -->
+  <PropertyGroup>
+    <!-- Foundry Local ships native runtime assets per RID, so restore must be RID-aware for the build to resolve them. -->
+    <RuntimeIdentifiers>win-x64;win-arm64;linux-x64;linux-arm64;osx-arm64</RuntimeIdentifiers>
+  </PropertyGroup>
+
+<!--#endif -->
   <ItemGroup>
 <!--#if (IsOllama && IsAspire)
     <PackageReference Include="CommunityToolkit.Aspire.OllamaSharp" Version="${PackageVersion:CommunityToolkit.Aspire.OllamaSharp}" />
 #endif -->
 <!--#if (IsOllama)
     <PackageReference Include="OllamaSharp" Version="${PackageVersion:OllamaSharp}" />
-#elif ((IsGHModels || IsOpenAI) && !IsAspire)
+#elif (IsOpenAI && !IsAspire)
+    <PackageReference Include="Microsoft.Extensions.AI.OpenAI" Version="${PackageVersion:Microsoft.Extensions.AI.OpenAI}" />
+#elif (IsFoundryLocal && !IsAspire)
+    <PackageReference Include="Microsoft.AI.Foundry.Local" Version="${PackageVersion:Microsoft.AI.Foundry.Local}" />
     <PackageReference Include="Microsoft.Extensions.AI.OpenAI" Version="${PackageVersion:Microsoft.Extensions.AI.OpenAI}" />
 #elif (IsAzureAIFoundry)
     <PackageReference Include="Azure.AI.Projects" Version="${PackageVersion:Azure.AI.Projects}" />
     <PackageReference Include="Microsoft.Extensions.AI.OpenAI" Version="${PackageVersion:Microsoft.Extensions.AI.OpenAI}" />
 #endif -->
-<!--#if ((IsAzureOpenAI || IsOpenAI || IsGHModels) && IsAspire) -->
+<!--#if ((IsAzureOpenAI || IsOpenAI) && IsAspire) -->
     <PackageReference Include="Aspire.Azure.AI.OpenAI" Version="${PackageVersion:Aspire.Azure.AI.OpenAI}" />
     <PackageReference Include="Microsoft.Extensions.AI.OpenAI" Version="${PackageVersion:Microsoft.Extensions.AI.OpenAI}" />
 <!--#endif -->

@@ -1,5 +1,9 @@
+#if (IsFoundryLocal)
+// Foundry Local has no .NET Aspire hosting integration and is only supported without Aspire.
+#error Foundry Local is not supported with .NET Aspire. Re-create the project without --aspire and with a non-Qdrant vector store (Qdrant requires Aspire).
+#endif
 using Microsoft.Extensions.AI;
-#if (IsOpenAI || IsGHModels)
+#if (IsOpenAI)
 using OpenAI;
 #endif
 using AIChatWeb_CSharp.Web.Components;
@@ -19,7 +23,7 @@
 builder.AddOllamaApiClient("embeddings")
     .AddEmbeddingGenerator();
 #elif (IsAzureAIFoundry)
-#else // (IsOpenAI || IsAzureOpenAI || IsGHModels)
+#else // (IsOpenAI || IsAzureOpenAI)
 #if (IsOpenAI)
 var openai = builder.AddOpenAIClient("openai");
 #else

@@ -1,8 +1,12 @@
-#if (IsGHModels || IsOpenAI || (IsAzureOpenAI && !IsManagedIdentity))
+#if (IsOpenAI || IsFoundryLocal || (IsAzureOpenAI && !IsManagedIdentity))
 using System.ClientModel;
 #elif (IsAzureOpenAI && IsManagedIdentity)
 using System.ClientModel.Primitives;
 #endif
+#if (IsFoundryLocal)
+using Microsoft.AI.Foundry.Local;
+using Microsoft.Extensions.Logging.Abstractions;
+#endif
 #if (IsAzureAISearch && !IsManagedIdentity)
 using Azure;
 #elif (IsManagedIdentity)
@@ -11,7 +15,7 @@
 using Microsoft.Extensions.AI;
 #if (IsOllama)
 using OllamaSharp;
-#elif (IsGHModels || IsOpenAI || IsAzureOpenAI)
+#elif (IsOpenAI || IsFoundryLocal || IsAzureOpenAI)
 using OpenAI;
 #endif
 using AIChatWeb_CSharp.Web.Components;
@@ -21,25 +25,52 @@
 var builder = WebApplication.CreateBuilder(args);
 builder.Services.AddRazorComponents().AddInteractiveServerComponents();
 
-#if (IsGHModels)
-// You will need to set the endpoint and key to your own values
-// You can do this using Visual Studio's "Manage User Secrets" UI, or on the command line:
-//   cd this-project-directory
-//   dotnet user-secrets set GitHubModels:Token YOUR-GITHUB-TOKEN
-var credential = new ApiKeyCredential(builder.Configuration["GitHubModels:Token"] ?? throw new InvalidOperationException("Missing configuration: GitHubModels:Token. See the README for details."));
-var openAIOptions = new OpenAIClientOptions()
-{
-    Endpoint = new Uri("https://models.inference.ai.azure.com")
-};
-
-var ghModelsClient = new OpenAIClient(credential, openAIOptions);
-var chatClient = ghModelsClient.GetChatClient("gpt-4o-mini").AsIChatClient();
-var embeddingGenerator = ghModelsClient.GetEmbeddingClient("text-embedding-3-small").AsIEmbeddingGenerator();
-#elif (IsOllama)
+#if (IsOllama)
 IChatClient chatClient = new OllamaApiClient(new Uri("http://localhost:11434"),
     "llama3.2");
 IEmbeddingGenerator<string, Embedding<float>> embeddingGenerator = new OllamaApiClient(new Uri("http://localhost:11434"),
     "all-minilm");
+#elif (IsFoundryLocal)
+var chatAlias = builder.Configuration["FoundryLocal:ChatModel"] ?? "qwen3-4b";
+var embeddingAlias = builder.Configuration["FoundryLocal:EmbeddingModel"] ?? "qwen3-embedding-0.6b";
+var foundryServiceUrl = builder.Configuration["FoundryLocal:ServiceUrl"] ?? "http://127.0.0.1:5273";
+
+await FoundryLocalManager.CreateAsync(new Configuration
+{
+    AppName = "AIChatWeb-CSharp",
+    Web = new Configuration.WebService { Urls = foundryServiceUrl }
+}, NullLogger.Instance);
+var foundryManager = FoundryLocalManager.Instance;
+await foundryManager.StartWebServiceAsync();
+var foundryCatalog = await foundryManager.GetCatalogAsync();
+
+async Task<string> EnsureFoundryModelAsync(string modelAlias)
+{
+    var model = await foundryCatalog.GetModelAsync(modelAlias)
+        ?? throw new InvalidOperationException(
+            $"Foundry Local model '{modelAlias}' was not found in the catalog. Run 'foundry model list' to see available models.");
+    if (!await model.IsCachedAsync())
+    {
+        Console.WriteLine($"Foundry Local: downloading model '{modelAlias}' (first run only)...");
+        await model.DownloadAsync(_ => { });
+    }
+    if (!await model.IsLoadedAsync())
+    {
+        await model.LoadAsync();
+    }
+    return model.Id;
+}
+
+var chatModelId = await EnsureFoundryModelAsync(chatAlias);
+var embeddingModelId = await EnsureFoundryModelAsync(embeddingAlias);
+
+var foundryEndpointUrl = foundryManager.Urls?.FirstOrDefault() ?? foundryServiceUrl;
+var foundryEndpoint = new Uri($"{foundryEndpointUrl.TrimEnd('/')}/v1");
+var foundryClient = new OpenAIClient(
+    new ApiKeyCredential("unused"),
+    new OpenAIClientOptions { Endpoint = foundryEndpoint });
+var chatClient = foundryClient.GetChatClient(chatModelId).AsIChatClient();
+var embeddingGenerator = foundryClient.GetEmbeddingClient(embeddingModelId).AsIEmbeddingGenerator();
 #elif (IsOpenAI)
 // You will need to set the endpoint and key to your own values
 // You can do this using Visual Studio's "Manage User Secrets" UI, or on the command line:

@@ -21,32 +21,6 @@ This incompatibility can be addressed by upgrading to Docker Desktop 4.41.1. See
 
 #### ---#endif
 # Configure the AI Model Provider
-#### ---#if (IsGHModels)
-To use models hosted by GitHub Models, you will need to create a GitHub personal access token with `models:read` permissions, but no other scopes or permissions. See [Prototyping with AI models](https://docs.github.com/github-models/prototyping-with-ai-models) and [Managing your personal access tokens](https://docs.github.com/authentication/keeping-your-account-and-data-secure/managing-your-personal-access-tokens) in the GitHub Docs for more information.
-
-#### ---#if (hostIdentifier == "vs")
-Configure your token for this project using .NET User Secrets:
-
-1. In Visual Studio, right-click on your project in the Solution Explorer and select "Manage User Secrets".
-2. This opens a `secrets.json` file where you can store your API keys without them being tracked in source control. Add the following key and value:
-
-   ```json
-   {
-     "GitHubModels:Token": "YOUR-TOKEN"
-   }
-   ```
-#### ---#else
-From the command line, configure your token for this project using .NET User Secrets by running the following commands:
-
-```sh
-cd <<your-project-directory>>
-dotnet user-secrets set GitHubModels:Token YOUR-TOKEN
-```
-#### ---#endif
-
-Learn more about [prototyping with AI models using GitHub Models](https://docs.github.com/github-models/prototyping-with-ai-models).
-
-#### ---#endif
 #### ---#if (IsOpenAI)
 ## Using OpenAI
 
@@ -92,6 +66,34 @@ ollama pull all-minilm
 ### 3. Learn more about Ollama
 Once the models are installed, you can start using them in your application. Refer to the [Ollama documentation](https://github.com/ollama/ollama/blob/main/docs/README.md) for detailed instructions on how to explore models locally.
 
+#### ---#endif
+#### ---#if (IsFoundryLocal)
+## Setting up a local environment using Foundry Local
+This project is configured to use Foundry Local, which runs models on your workstation through a local OpenAI-compatible endpoint. It does not need an API key.
+
+### 1. Install Foundry Local
+Install Foundry Local for your operating system by following the [Foundry Local documentation](https://learn.microsoft.com/azure/ai-foundry/foundry-local/).
+
+### 2. Run the app
+The app starts the Foundry Local service for you. On first run, it downloads the configured models, then loads them into the local service.
+
+The default chat model alias is `qwen3-4b`. The default embedding model alias is `qwen3-embedding-0.6b`.
+
+### 3. Override model aliases or the service URL
+You can change the defaults in `appsettings.json`, `appsettings.Development.json`, or user secrets:
+
+```json
+{
+  "FoundryLocal": {
+    "ChatModel": "qwen3-4b",
+    "EmbeddingModel": "qwen3-embedding-0.6b",
+    "ServiceUrl": "http://127.0.0.1:5273"
+  }
+}
+```
+
+Use `FoundryLocal:ServiceUrl` if another local process already uses the default port.
+
 #### ---#endif
 #### ---#if (IsAzureOpenAI)
 ## Using Azure OpenAI

@@ -7,6 +7,8 @@ public class IngestedChunk
 {
 #if (IsOllama)
     public const int VectorDimensions = 384; // 384 is the default vector size for the all-minilm embedding model
+#elif (IsFoundryLocal)
+    public const int VectorDimensions = 1024; // 1024 is the default vector size for the qwen3-embedding-0.6b embedding model
 #else
     public const int VectorDimensions = 1536; // 1536 is the default vector size for the OpenAI text-embedding-3-small model
 #endif
@@ -17,9 +19,15 @@ public class IngestedChunk
 #endif
     public const string CollectionName = "data-AIChatWeb-CSharp.Web-chunks";
 
+#if (IsQdrant)
     [VectorStoreKey(StorageName = "key")]
     [JsonPropertyName("key")]
     public required Guid Key { get; set; }
+#else
+    [VectorStoreKey(StorageName = "key")]
+    [JsonPropertyName("key")]
+    public required string Key { get; set; }
+#endif
 
     [VectorStoreData(StorageName = "documentid")]
     [JsonPropertyName("documentid")]

@@ -6,4 +6,11 @@
       "Microsoft.EntityFrameworkCore": "Warning"
     }
   }
+//#if (IsFoundryLocal)
+  ,"FoundryLocal": {
+    "ChatModel": "qwen3-4b",
+    "EmbeddingModel": "qwen3-embedding-0.6b",
+    "ServiceUrl": "http://127.0.0.1:5273"
+  }
+//#endif
 }
@@ -6,5 +6,12 @@
       "Microsoft.EntityFrameworkCore": "Warning"
     }
   },
+//#if (IsFoundryLocal)
+  "FoundryLocal": {
+    "ChatModel": "qwen3-4b",
+    "EmbeddingModel": "qwen3-embedding-0.6b",
+    "ServiceUrl": "http://127.0.0.1:5273"
+  },
+//#endif
   "AllowedHosts": "*"
 }