Pinecone .NET Library

The official Pinecone .NET library supporting .NET Standard, .NET Core, and .NET Framework.

Requirements

To use this SDK, ensure that your project is targeting one of the following:

• .NET Standard 2.0+
• .NET Core 3.0+
• .NET Framework 4.6.2+
• .NET 6.0+

Installation

Using the .NET Core command-line interface (CLI) tools:

dotnet add package Pinecone.Client

Using the NuGet Command Line Interface (CLI):

nuget install Pinecone.Client

Documentation

API reference documentation is available here.

Usage

Instantiate the SDK using the Pinecone class.

using Pinecone;

var pinecone = new PineconeClient("PINECONE_API_KEY")

Indexes

Operations related to the building and managing of Pinecone indexes are called control plane operations.

Create index

You can use the .NET SDK to create two types of indexes:

1. Serverless indexes (recommended for most use cases)
2. Pod-based indexes (recommended for high-throughput use cases).

Create a serverless index

The following is an example of creating a serverless index in the us-east-1 region of AWS. For more information on serverless and regional availability, see Understanding indexes.

using Pinecone;

var pinecone = new PineconeClient("PINECONE_API_KEY");

var index = await pinecone.CreateIndexAsync(new CreateIndexRequest
{
   Name = "example-index",
   Dimension = 1538,
   Metric = CreateIndexRequestMetric.Cosine,
   Spec = new ServerlessIndexSpec
   {
       Serverless = new ServerlessSpec
       {
           Cloud = ServerlessSpecCloud.Azure,
           Region = "eastus2",
       }
   },
   DeletionProtection = DeletionProtection.Enabled
});

Create a pod-based index

The following is a minimal example of creating a pod-based index.

using Pinecone;

var pinecone = new PineconeClient("PINECONE_API_KEY");

var index = await pinecone.CreateIndexAsync(new CreateIndexRequest
{
   Name = "example-index",
   Dimension = 1538,
   Metric = CreateIndexRequestMetric.Cosine,
   Spec = new PodIndexSpec
   {
       Pod = new PodSpec
       {
           Environment = "eastus-azure",
           PodType = "p1.x1",
           Pods = 1,
           Replicas = 1,
           Shards = 1,
       }
   },
   DeletionProtection = DeletionProtection.Enabled
});

List indexes

The following example returns all indexes (and their corresponding metadata) in your project.

using Pinecone;

var pinecone = new PineconeClient("PINECONE_API_KEY");

var indexesInYourProject = await pinecone.ListIndexesAsync();

Delete an index

The following example deletes an index by name.

using Pinecone;

var pinecone = new PineconeClient("PINECONE_API_KEY");

await pinecone.DeleteIndexAsync("example-index");

Describe an index

The following example returns metadata about an index.

using Pinecone;

var pinecone = new PineconeClient("PINECONE_API_KEY");

var indexModel = pinecone.DescribeIndexAsync("example-index");

Scale replicas

The following example changes the number of replicas for an index.

using Pinecone;

var pinecone = new PineconeClient("PINECONE_API_KEY");

var indexMetadata = await pinecone.ConfigureIndexAsync("example-index", new ConfigureIndexRequest
{
   Spec = new ConfigureIndexRequestSpec
   {
       Pod = new ConfigureIndexRequestSpecPod
       {
           Replicas = 2,
           PodType = "p1.x1",
       }
   }
});

Note that scaling replicas is only applicable to pod-based indexes.

Describe index statistics

The following example returns statistics about an index.

using Pinecone;

var pinecone = new PineconeClient("PINECONE_API_KEY");

var index = pinecone.Index("example-index");
var indexStatsResponse = await index.DescribeIndexStatsAsync(new DescribeIndexStatsRequest());

Upsert vectors

Operations related to the indexing, deleting, and querying of vectors are called data plane operations.

The following example upserts vectors to example-index.

using Pinecone;

var pinecone = new PineconeClient("PINECONE_API_KEY");

var index = pinecone.Index("example-index");

// Vector ids to be upserted
var upsertIds = new[] { "v1", "v2", "v3" };

// List of values to be upserted
float[][] values =
[
    [1.0f, 2.0f, 3.0f],
    [4.0f, 5.0f, 6.0f],
    [7.0f, 8.0f, 9.0f],
];

// List of sparse indices to be upserted
uint[][] sparseIndices =
[
    [1, 2, 3],
    [4, 5, 6],
    [7, 8, 9],
];

// List of sparse values to be upserted
float[][] sparseValues =
[
    [1000f, 2000f, 3000f],
    [4000f, 5000f, 6000f],
    [7000f, 8000f, 9000f],
];

// Metadata to be upserted
var metadataStructArray = new[]
{
    new Metadata { ["genre"] = "action", ["year"] = 2019 },
    new Metadata { ["genre"] = "thriller", ["year"] = 2020 },
    new Metadata { ["genre"] = "comedy", ["year"] = 2021 },
};

var vectors = new List<Vector>();
for (var i = 0; i <= 2; i++)
{
    vectors.Add(
        new Vector
        {
            Id = upsertIds[i],
            Values = values[i],
            SparseValues = new SparseValues
            {
                Indices = sparseIndices[i],
                Values = sparseValues[i],
            },
            Metadata = metadataStructArray[i],
        }
    );
}

var upsertResponse = await index.UpsertAsync(new UpsertRequest { Vectors = vectors, });

Query an index

The following example queries the index example-index with metadata filtering.

using Pinecone;

var pinecone = new PineconeClient("PINECONE_API_KEY");

var index = pinecone.Index("example-index");

var queryResponse = await index.QueryAsync(
   new QueryRequest
   {
       Namespace = "example-namespace",
       Vector = [0.1f, 0.2f, 0.3f, 0.4f],
       TopK = 10,
       IncludeValues = true,
       IncludeMetadata = true,
       Filter = new Metadata
       {
           ["genre"] =
               new Metadata
               {
                   ["$in"] = new[] { "comedy", "documentary", "drama" },
               }
       }
   });

Query sparse-dense vectors

The following example queries an index using a sparse-dense vector:

using Pinecone;
var pinecone = new PineconeClient("PINECONE_API_KEY");

var index = pinecone.Index("example-index");

var queryResponse = await index.QueryAsync(
   new QueryRequest
   {
       TopK = 10,
       Vector = [0.1f, 0.2f, 0.3f],
       SparseVector = new SparseValues
       {
           Indices = [10, 45, 16],
           Values = [0.5f, 0.5f, 0.2f],
       }
   }
);

Delete vectors

The following example deletes vectors by ID.

using Pinecone;

var pinecone = new PineconeClient("PINECONE_API_KEY");

var index = pinecone.Index("example-index");

var deleteResponse = await index.DeleteAsync(new DeleteRequest
{
   Ids = new[] { "v1" },
   Namespace = "example-namespace",
});

The following example deletes all records in a namespace and the namespace itself:

using Pinecone;

var pinecone = new PineconeClient("PINECONE_API_KEY");

var index = pinecone.Index("example-index");

var deleteResponse = await index.DeleteAsync(new DeleteRequest {
    DeleteAll = true,
    Namespace = "example-namespace",
});

Fetch vectors

The following example fetches vectors by ID.

using Pinecone;

var pinecone = new PineconeClient("PINECONE_API_KEY");

var index = pinecone.Index("example-index");

var fetchResponse = await index.FetchAsync(new FetchRequest {
    Ids = new[] { "v1" },
    Namespace = "example-namespace",
});

List vector IDs

The following example lists up to 100 vector IDs from a Pinecone index.

The following demonstrates how to use the list endpoint to get vector IDs from a specific namespace, filtered by a given prefix.

using Pinecone;

var pinecone = new PineconeClient("PINECONE_API_KEY");

var index = pinecone.Index("example-index");

var listResponse = await index.ListAsync(new ListRequest {
    Namespace = "example-namespace",
    Prefix = "prefix-",
});

Update vectors

The following example updates vectors by ID.

using Pinecone;

var pinecone = new PineconeClient("PINECONE_API_KEY");

var index = pinecone.Index("example-index");

var updateResponse = await index.UpdateAsync(new UpdateRequest
{
   Id = "vec1",
   Values = new[] { 0.1f, 0.2f, 0.3f, 0.4f },
   SetMetadata = new Metadata { ["genre"] = "drama" },
   Namespace = "example-namespace",
});

Collections

Collections fall under data plane operations.

Create a collection

The following creates a collection.

using Pinecone;

var pinecone = new PineconeClient("PINECONE_API_KEY");

var collectionModel = await pinecone.CreateCollectionAsync(new CreateCollectionRequest {
    Name = "example-collection",
    Source = "example-index",
});

List collections

The following example returns a list of the collections in the current project.

using Pinecone;

var pinecone = new PineconeClient("PINECONE_API_KEY");

var collectionList = await pinecone.ListCollectionsAsync();

Describe a collection

The following example returns a description of the collection.

using Pinecone;

var pinecone = new PineconeClient("PINECONE_API_KEY");

var collectionModel = await pinecone.DescribeCollectionAsync("example-collection");

Delete a collection

The following example deletes the collection example-collection.

using Pinecone;

var pinecone = new PineconeClient("PINECONE_API_KEY");

await pinecone.DeleteCollectionAsync("example-collection");

Inference

Embed

The Pinecone SDK now supports creating embeddings via the Inference API.

using Pinecone;

var pinecone = new PineconeClient("PINECONE_API_KEY");

// Prepare input sentences to be embedded
List<EmbedRequestInputsItem> inputs =
[
    new()
    {
        Text = "The quick brown fox jumps over the lazy dog."
    },
    new()
    {
        Text = "Lorem ipsum"
    }
];

// Specify the embedding model and parameters
var embeddingModel = "multilingual-e5-large";

// Generate embeddings for the input data
var embeddings = await pinecone.Inference.EmbedAsync(new EmbedRequest()
{
    Model = embeddingModel,
    Inputs = inputs,
    Parameters = new EmbedRequestParameters()
    {
        InputType = "query",
        Truncate = "END"
    }
});

// Get embedded data
var embeddedData = embeddings.Data;

Rerank

The following example shows how to rerank items according to their relevance to a query.

using Pinecone;

var pinecone = new PineconeClient("PINECONE_API_KEY");

// The model to use for reranking
var model = "bge-reranker-v2-m3";

// The query to rerank documents against
var query = "The tech company Apple is known for its innovative products like the iPhone.";

// Add the documents to rerank
var documents = new List<Dictionary<string, string>>
{
    new()
    {
        ["id"] = "vec1",
        ["my_field"] = "Apple is a popular fruit known for its sweetness and crisp texture."
    },
    new()
    {
        ["id"] = "vec2",
        ["my_field"] = "Many people enjoy eating apples as a healthy snack."
    },
    new()
    {
        ["id"] = "vec3",
        ["my_field"] =
            "Apple Inc. has revolutionized the tech industry with its sleek designs and user-friendly interfaces."
    },
    new()
    {
        ["id"] = "vec4",
        ["my_field"] = "An apple a day keeps the doctor away, as the saying goes."
    }
};

// The fields to rank the documents by. If not provided, the default is "text"
var rankFields = new List<string> { "my_field" };

// The number of results to return sorted by relevance. Defaults to the number of inputs
int topN = 2;

// Whether to return the documents in the response
bool returnDocuments = true;

// Additional model-specific parameters for the reranker
var parameters = new Dictionary<string, string>
{
    ["truncate"] = "END"
};

// Send ranking request
var result = await pinecone.Inference.RerankAsync(
    new RerankRequest
    {
        Model = model,
        Query = query,
        Documents = documents,
        RankFields = rankFields,
        TopN = topN,
        Parameters = parameters
    });

// Get ranked data
var data = result.Data;

Imports

Start an import

The following example initiates an asynchronous import of vectors from object storage into the index.

using Pinecone;

var pinecone = new PineconeClient("PINECONE_API_KEY");
var index = pinecone.Index("PINECONE_INDEX_NAME");

var uri = "s3://path/to/file.parquet";

var response = await index.StartBulkImportAsync(new StartImportRequest
{
    Uri = uri,
    IntegrationId = "123-456-789",
    ErrorMode = new ImportErrorMode { OnError = ImportErrorModeOnError.Continue }
});

List imports

The following example lists all recent and ongoing import operations for the specified index.

using Pinecone;

var pinecone = new PineconeClient("PINECONE_API_KEY");

var index = pinecone.Index("PINECONE_INDEX_NAME");

var imports = await index.ListBulkImportsAsync(new ListBulkImportsRequest
{
    Limit = 100,
    PaginationToken = "some-pagination-token"
});

Describe an import

The following example retrieves detailed information about a specific import operation using its unique identifier.

using Pinecone;

var pinecone = new PineconeClient("PINECONE_API_KEY");

var index = pinecone.Index("PINECONE_INDEX_NAME");

var importDetails = await index.DescribeBulkImportAsync("1");

Cancel an import

The following example attempts to cancel an ongoing import operation using its unique identifier.

using Pinecone;

var pinecone = new PineconeClient("PINECONE_API_KEY");

var index = pinecone.Index("PINECONE_INDEX_NAME");

var cancelResponse = await index.CancelBulkImportAsync("2");

Advanced

Control Plane Client Options

Control Plane endpoints are accessed via standard HTTP requests. You can configure the following HTTP client options:

• MaxRetries: The maximum number of times the client will retry a failed request. Default is 2.
• Timeout: The time limit for each request before it times out. Default is 30 seconds.
• BaseUrl: The base URL for all requests.
• HttpClient: The HTTP client to be used for all requests.
• IsTlsEnabled: The client will default to using HTTPS if true, and to HTTP if false. Default is true.

Example usage:

var pinecone = new PineconeClient("PINECONE_API_KEY", new ClientOptions
{
    MaxRetries = 3,
    Timeout = TimeSpan.FromSeconds(60),
    HttpClient = ..., // Override the Http Client
    BaseUrl = ..., // Override the Base URL
    IsTlsEnabled = true
});

Configuring HTTP proxy for both control and data plane operations

If your network setup requires you to interact with Pinecone via a proxy, you need to configure the HTTP client accordingly.

using System.Net;
using Pinecone;

var pinecone = new PineconeClient("PINECONE_API_KEY", new ClientOptions
{
    HttpClient = new HttpClient(new HttpClientHandler
    {
        Proxy = new WebProxy("PROXY_HOST:PROXY_PORT")
    })
});

If you're building your HTTP client using the HTTP client factory, you can use the ConfigurePrimaryHttpMessageHandler method to configure the proxy.

   .ConfigurePrimaryHttpMessageHandler(() => new HttpClientHandler
       {
           Proxy = new WebProxy("PROXY_HOST:PROXY_PORT")
       });

Data Plane gRPC Options

Data Plane endpoints are accessed via gRPC. You can configure the Pinecone client with gRPC channel options for advanced control over gRPC communication settings. These options allow you to customize various aspects like message size limits, retry attempts, credentials, and more.

Example usage:

var pinecone = new PineconeClient("PINECONE_API_KEY", new ClientOptions
{
    GrpcOptions = new GrpcChannelOptions
    {
        MaxRetryAttempts = 5,
        MaxReceiveMessageSize = 4 * 1024 * 1024 // 4 MB
        // Additional configuration options...
    }
});

Exception handling

When the API returns a non-zero status code, (4xx or 5xx response), a subclass of PineconeException will be thrown:

try {
    pinecone.CreateIndexAsync(...);
} catch (PineconeException e) {
    System.Console.WriteLine(e.Message)
    System.Console.WriteLine(e.StatusCode)
}

Contributing

While we value open-source contributions to this SDK, this library is generated programmatically. Additions made directly to this library would have to be moved over to our generation code, otherwise they would be overwritten upon the next generated release. Feel free to open a PR as a proof of concept, but know that we will not be able to Pinecone it as-is. We suggest opening an issue first to discuss with us!

On the other hand, contributions to the README are always very welcome!