AspireAzureAIInferenceExtensions Methods

ClassMethods8 members

Extension methods for adding Azure AI Inference services to an Aspire application.

AddAzureChatCompletionsClient(IHostApplicationBuilder, string, Action<ChatCompletionsClientSettings>, Action<IAzureClientBuilder<ChatCompletionsClient, AzureAIInferenceClientOptions>>)

extensionAspireChatCompletionsClientBuilder

Adds a Inference.ChatCompletionsClient to the application and configures it with the specified settings.

public static class AspireAzureAIInferenceExtensions
{
    public static AspireChatCompletionsClientBuilder AddAzureChatCompletionsClient(
        this IHostApplicationBuilder builder,
        string connectionName,
        Action<ChatCompletionsClientSettings>? configureSettings = null,
        Action<IAzureClientBuilder<ChatCompletionsClient, AzureAIInferenceClientOptions>>? configureClientBuilder = null)
    {
        // ...
    }
}

AspireAzureAIInferenceExtensions.cs#L65-L76

Parameters

builderIHostApplicationBuilderThe Hosting.IHostApplicationBuilder to add the client to.

connectionNamestringThe name of the client. This is used to retrieve the connection string from configuration.

configureSettingsAction<ChatCompletionsClientSettings>optionalAn optional callback to configure the ChatCompletionsClientSettings.

configureClientBuilderAction<IAzureClientBuilder<ChatCompletionsClient, AzureAIInferenceClientOptions>>optionalAn optional callback to configure the Extensions.IAzureClientBuilder`2 for the client.

Returns

AspireChatCompletionsClientBuilderAn AspireChatCompletionsClientBuilder that can be used to further configure the client.

Exceptions

InvalidOperationExceptionThrown when endpoint is missing from settings.

Remarks

The client is registered as a singleton.

Configuration is loaded from the "Aspire:Azure:AI:Inference" section, and can be supplemented with a connection string named after the connectionName parameter.

Examples

The following example shows how to register a and use it in a minimal API endpoint:

var builder = WebApplication.CreateBuilder(args);
builder.AddAzureChatCompletionsClient("chat");

var app = builder.Build();
app.MapGet("/chat", async (ChatCompletionsClient client) =>
{
    var response = await client.CompleteAsync("Hello, how are you?");
    return response.Value.Content;
});

AddAzureEmbeddingsClient(IHostApplicationBuilder, string, Action<ChatCompletionsClientSettings>, Action<IAzureClientBuilder<EmbeddingsClient, AzureAIInferenceClientOptions>>)

extensionAspireEmbeddingsClientBuilder

Adds a Inference.EmbeddingsClient to the application and configures it with the specified settings.

public static class AspireAzureAIInferenceExtensions
{
    public static AspireEmbeddingsClientBuilder AddAzureEmbeddingsClient(
        this IHostApplicationBuilder builder,
        string connectionName,
        Action<ChatCompletionsClientSettings>? configureSettings = null,
        Action<IAzureClientBuilder<EmbeddingsClient, AzureAIInferenceClientOptions>>? configureClientBuilder = null)
    {
        // ...
    }
}

AspireAzureAIInferenceExtensions.cs#L321-L332

Parameters

builderIHostApplicationBuilderThe Hosting.IHostApplicationBuilder to add the client to.

connectionNamestringThe name of the client. This is used to retrieve the connection string from configuration.

configureSettingsAction<ChatCompletionsClientSettings>optionalAn optional callback to configure the ChatCompletionsClientSettings.

configureClientBuilderAction<IAzureClientBuilder<EmbeddingsClient, AzureAIInferenceClientOptions>>optionalAn optional callback to configure the Extensions.IAzureClientBuilder`2 for the client.

Returns

AspireEmbeddingsClientBuilderAn AspireEmbeddingsClientBuilder that can be used to further configure the client.

Exceptions

InvalidOperationExceptionThrown when endpoint is missing from settings.

Remarks

The client is registered as a singleton.

Configuration is loaded from the "Aspire:Azure:AI:Inference" section, and can be supplemented with a connection string named after the connectionName parameter.

Examples

The following example shows how to register an and use it in a minimal API endpoint:

var builder = WebApplication.CreateBuilder(args);
builder.AddAzureEmbeddingsClient("embeddings");

var app = builder.Build();
app.MapGet("/embed", async (EmbeddingsClient client) =>
{
    var response = await client.EmbedAsync("Hello, world!");
    return response.Value.Data;
});

AddChatClient(AspireChatCompletionsClientBuilder, string?)extensionChatClientBuilder

Creates a AI.IChatClient from the Inference.ChatCompletionsClient registered in the service collection.

public static class AspireAzureAIInferenceExtensions
{
    public static ChatClientBuilder AddChatClient(
        this AspireChatCompletionsClientBuilder builder,
        string? deploymentName = null)
    {
        // ...
    }
}

AspireAzureAIInferenceExtensions.cs#L226-L229

Parameters

builderAspireChatCompletionsClientBuilderAn AspireChatCompletionsClientBuilder.

deploymentNamestring?optionalOptionally specifies which model deployment to use. If not specified, a value will be taken from the connection string.

Returns

ChatClientBuilderA AI.ChatClientBuilder.

Examples

The following example shows how to register an and use it in a minimal API endpoint:

var builder = WebApplication.CreateBuilder(args);
builder.AddAzureChatCompletionsClient("chat").AddChatClient();

var app = builder.Build();
app.MapGet("/chat", async (IChatClient client) =>
{
    var response = await client.GetResponseAsync("Hello, how are you?");
    return response.Text;
});

AddEmbeddingGenerator(AspireEmbeddingsClientBuilder, string?)extensionEmbeddingGeneratorBuilder<string, Embedding<float>>

Creates a AI.IEmbeddingGenerator`2 from the Inference.EmbeddingsClient registered in the service collection.

public static class AspireAzureAIInferenceExtensions
{
    public static EmbeddingGeneratorBuilder<string, Embedding<float>> AddEmbeddingGenerator(
        this AspireEmbeddingsClientBuilder builder,
        string? deploymentName = null)
    {
        // ...
    }
}

AspireAzureAIInferenceExtensions.cs#L479-L482

Parameters

builderAspireEmbeddingsClientBuilderAn AspireEmbeddingsClientBuilder.

deploymentNamestring?optionalOptionally specifies which model deployment to use. If not specified, a value will be taken from the connection string.

Returns

EmbeddingGeneratorBuilder<string, Embedding<float>>An AI.EmbeddingGeneratorBuilder`2.

Examples

The following example shows how to register an and use it in a minimal API endpoint:

var builder = WebApplication.CreateBuilder(args);
builder.AddAzureEmbeddingsClient("embeddings").AddEmbeddingGenerator();

var app = builder.Build();
app.MapGet("/embed", async (IEmbeddingGenerator<string, Embedding<float>> generator) =>
{
    var embedding = await generator.GenerateEmbeddingAsync("Hello, world!");
    return embedding.Vector.ToArray();
});

AddKeyedAzureChatCompletionsClient(IHostApplicationBuilder, string, Action<ChatCompletionsClientSettings>, Action<IAzureClientBuilder<ChatCompletionsClient, AzureAIInferenceClientOptions>>)

extensionAspireChatCompletionsClientBuilder

Adds a Inference.ChatCompletionsClient to the application and configures it with the specified settings.

public static class AspireAzureAIInferenceExtensions
{
    public static AspireChatCompletionsClientBuilder AddKeyedAzureChatCompletionsClient(
        this IHostApplicationBuilder builder,
        string name,
        Action<ChatCompletionsClientSettings>? configureSettings = null,
        Action<IAzureClientBuilder<ChatCompletionsClient, AzureAIInferenceClientOptions>>? configureClientBuilder = null)
    {
        // ...
    }
}

AspireAzureAIInferenceExtensions.cs#L116-L127

Parameters

builderIHostApplicationBuilderThe Hosting.IHostApplicationBuilder to add the client to.

namestringThe name of the component, which is used as the ServiceDescriptor.ServiceKey of the service and also to retrieve the connection string from the ConnectionStrings configuration section.

configureSettingsAction<ChatCompletionsClientSettings>optionalAn optional callback to configure the ChatCompletionsClientSettings.

configureClientBuilderAction<IAzureClientBuilder<ChatCompletionsClient, AzureAIInferenceClientOptions>>optionalAn optional callback to configure the Extensions.IAzureClientBuilder`2 for the client.

Returns

AspireChatCompletionsClientBuilderAn AspireChatCompletionsClientBuilder that can be used to further configure the client.

Exceptions

InvalidOperationExceptionThrown when endpoint is missing from settings.

Remarks

The client is registered as a singleton with a keyed service.

Configuration is loaded from the "Aspire:Azure:AI:Inference" section, and can be supplemented with a connection string named after the name parameter.

Examples

The following example shows how to register a keyed and use it in a minimal API endpoint:

var builder = WebApplication.CreateBuilder(args);
builder.AddKeyedAzureChatCompletionsClient("chat");

var app = builder.Build();
app.MapGet("/chat", async ([FromKeyedServices("chat")] ChatCompletionsClient client) =>
{
    var response = await client.CompleteAsync("Hello, how are you?");
    return response.Value.Content;
});

AddKeyedAzureEmbeddingsClient(IHostApplicationBuilder, string, Action<ChatCompletionsClientSettings>, Action<IAzureClientBuilder<EmbeddingsClient, AzureAIInferenceClientOptions>>)

extensionAspireEmbeddingsClientBuilder

Adds a Inference.EmbeddingsClient to the application and configures it with the specified settings.

public static class AspireAzureAIInferenceExtensions
{
    public static AspireEmbeddingsClientBuilder AddKeyedAzureEmbeddingsClient(
        this IHostApplicationBuilder builder,
        string name,
        Action<ChatCompletionsClientSettings>? configureSettings = null,
        Action<IAzureClientBuilder<EmbeddingsClient, AzureAIInferenceClientOptions>>? configureClientBuilder = null)
    {
        // ...
    }
}

AspireAzureAIInferenceExtensions.cs#L372-L383

Parameters

builderIHostApplicationBuilderThe Hosting.IHostApplicationBuilder to add the client to.

namestringThe name of the component, which is used as the ServiceDescriptor.ServiceKey of the service and also to retrieve the connection string from the ConnectionStrings configuration section.

configureSettingsAction<ChatCompletionsClientSettings>optionalAn optional callback to configure the ChatCompletionsClientSettings.

configureClientBuilderAction<IAzureClientBuilder<EmbeddingsClient, AzureAIInferenceClientOptions>>optionalAn optional callback to configure the Extensions.IAzureClientBuilder`2 for the client.

Returns

AspireEmbeddingsClientBuilderAn AspireEmbeddingsClientBuilder that can be used to further configure the client.

Exceptions

InvalidOperationExceptionThrown when endpoint is missing from settings.

Remarks

The client is registered as a singleton with a keyed service.

Configuration is loaded from the "Aspire:Azure:AI:Inference" section, and can be supplemented with a connection string named after the name parameter.

Examples

The following example shows how to register a keyed and use it in a minimal API endpoint:

var builder = WebApplication.CreateBuilder(args);
builder.AddKeyedAzureEmbeddingsClient("embeddings");

var app = builder.Build();
app.MapGet("/embed", async ([FromKeyedServices("embeddings")] EmbeddingsClient client) =>
{
    var response = await client.EmbedAsync("Hello, world!");
    return response.Value.Data;
});

AddKeyedChatClient(AspireChatCompletionsClientBuilder, string, string?)extensionChatClientBuilder

Creates a AI.IChatClient from the Inference.ChatCompletionsClient registered in the service collection.

public static class AspireAzureAIInferenceExtensions
{
    public static ChatClientBuilder AddKeyedChatClient(
        this AspireChatCompletionsClientBuilder builder,
        string serviceKey,
        string? deploymentName = null)
    {
        // ...
    }
}

AspireAzureAIInferenceExtensions.cs#L255-L261

Parameters

builderAspireChatCompletionsClientBuilderAn AspireChatCompletionsClientBuilder.

serviceKeystringThe service key with which the AI.IChatClient will be registered.

deploymentNamestring?optionalOptionally specifies which model deployment to use. If not specified, a value will be taken from the connection string.

Returns

ChatClientBuilderA AI.ChatClientBuilder.

Examples

The following example shows how to register a keyed and use it in a minimal API endpoint:

var builder = WebApplication.CreateBuilder(args);
builder.AddKeyedAzureChatCompletionsClient("chat").AddKeyedChatClient("chat");

var app = builder.Build();
app.MapGet("/chat", async ([FromKeyedServices("chat")] IChatClient client) =>
{
    var response = await client.GetResponseAsync("Hello, how are you?");
    return response.Text;
});

AddKeyedEmbeddingGenerator(AspireEmbeddingsClientBuilder, string, string?)extensionEmbeddingGeneratorBuilder<string, Embedding<float>>

Creates a AI.IEmbeddingGenerator`2 from the Inference.EmbeddingsClient registered in the service collection.

public static class AspireAzureAIInferenceExtensions
{
    public static EmbeddingGeneratorBuilder<string, Embedding<float>> AddKeyedEmbeddingGenerator(
        this AspireEmbeddingsClientBuilder builder,
        string serviceKey,
        string? deploymentName = null)
    {
        // ...
    }
}

AspireAzureAIInferenceExtensions.cs#L508-L514

Parameters

builderAspireEmbeddingsClientBuilderAn AspireEmbeddingsClientBuilder.

serviceKeystringThe service key with which the AI.IEmbeddingGenerator`2 will be registered.

deploymentNamestring?optionalOptionally specifies which model deployment to use. If not specified, a value will be taken from the connection string.

Returns

EmbeddingGeneratorBuilder<string, Embedding<float>>An AI.EmbeddingGeneratorBuilder`2.

Examples

The following example shows how to register a keyed and use it in a minimal API endpoint:

var builder = WebApplication.CreateBuilder(args);
builder.AddKeyedAzureEmbeddingsClient("embeddings").AddKeyedEmbeddingGenerator("embeddings");

var app = builder.Build();
app.MapGet("/embed", async ([FromKeyedServices("embeddings")] IEmbeddingGenerator<string, Embedding<float>> generator) =>
{
    var embedding = await generator.GenerateEmbeddingAsync("Hello, world!");
    return embedding.Vector.ToArray();
});