Google.GenAI 0.4.0

Google Gen AI .Net SDK provides an interface for developers to integrate Google's generative models into their .Net applications. It supports the Gemini Developer API and Vertex AI APIs.

Supported .NET version

This library is built for .NET 8.0 and netstandard2.1.

Full API Reference

The full API reference is hosted in the dedicated GitHub Page

Install

In your dotnet project directory, type the the following command

dotnet add package Google.GenAI

Imports

using Google.GenAI;
using Google.GenAI.Types;

Create a client

Please run one of the following code blocks to create a client for different services (Gemini Developer API or Vertex AI).

using Google.GenAI;

//  Only run this block for Gemini Developer API
var client = new Client(apiKey: apiKey);

using Google.GenAI;

// only run this block for Vertex AI API
client = new Client(
    project: project, location: location, vertexAI: true
)

(Optional) Using environment variables:

You can create a client by configuring the necessary environment variables. Configuration setup instructions depends on whether you're using the Gemini Developer API or the Gemini API in Vertex AI.

Gemini Developer API: Set the GOOGLE_API_KEY. It will automatically be picked up by the client.

export GEMINI_API_KEY='your-api-key'

Gemini API on Vertex AI: Set GOOGLE_GENAI_USE_VERTEXAI, GOOGLE_CLOUD_PROJECT and GOOGLE_CLOUD_LOCATION, as shown below:

export GOOGLE_GENAI_USE_VERTEXAI=true
export GOOGLE_CLOUD_PROJECT='your-project-id'
export GOOGLE_CLOUD_LOCATION='us-central1'

using Google.GenAI;

client = new Client();

Types

Parameter types are specified in the Google.GenAI.Types namespace.

Models

The client.Models module exposes model inferencing. See Create a client section above to initialize a client.

Generate Content

With simple text content

using System.Threading.Tasks;
using Google.GenAI;
using Google.GenAI.Types;

public class GenerateContentSimpleText {
  public static async Task main() {
    // assuming credentials are set up in environment variables as instructed above.
    var client = new Client();
    var response = await client.Models.GenerateContentAsync(
      model: "gemini-2.0-flash", contents: "why is the sky blue?"
    );
    Console.WriteLine(response.Candidates[0].Content.Parts[0].Text);
  }
}

System Instructions and Other Configs

The output of the model can be influenced by several optional settings available in GenerateContentAsync's config parameter. For example, to make a model more deterministic, lowering the Temperature parameter reduces randomness, with values near 0 minimizing variability. Capabilities and parameter defaults for each model is shown in the Vertex AI docs and Gemini API docs respectively.

using System.Threading.Tasks;
using Google.GenAI;
using Google.GenAI.Types;

public class GenerateContentWithConfig {
  public static async Task main() {
    // assuming credentials are set up in environment variables as instructed above.
    var client = new Client();
    var generateContentConfig =
    new GenerateContentConfig {
        SystemInstruction = new Content
        {
          Parts = new List<Part> {
              new Part {Text = "I say high you say low."}
          }
        },
        Temperature = 0.1,
        MaxOutputTokens = 3
    };
    var response = await client.Models.GenerateContentAsync(
        model: "gemini-2.0-flash",
        contents: "high",
        config: generateContentConfig
    );
    Console.WriteLine(response.Candidates[0].Content.Parts[0].Text);
  }
}

Safety Settings

using System.Threading.Tasks;
using Google.GenAI;
using Googel.GenAI.Types;

class GenerateContentWithSafetySettings {
  public static async Task main() {
    // assuming credentials are set up in environment variables as instructed above.
    var client = new Client();

    var safetySettings = new List<SafetySetting> {
      new SafetySetting {
        Category = HarmCategory.HARM_CATEGORY_HATE_SPEECH,
        Threshold = HarmBlockThreshold.BLOCK_LOW_AND_ABOVE
      }
    };
    var generateContentConfig = new GenerateContentConfig
    {
      SafetySettings =  new List<SafetySetting>(safetySettings)
    };
    var response = await client.Models.GenerateContentAsync(
        model: "gemini-2.0-flash",
        contents: "say something hateful",
        config: generateContentConfig
    );
    Console.WriteLine(response.Candidates[0].SafetyRatings);
  }
}

Json response scehema

However you define your schema, don't duplicate it in your input prompt, including by giving examples of expected JSON output. If you do, the generated output might be lower in quality.

using System.Threading.Tasks;
using Goolge.GenAI;
using Google.GenAI.Types;

public class GenerateContentWithJsonSchema {
  public static async Task main() {
    // assuming credentials are set up in environment variables as instructed above.
    var client = new Client();

    // define the response schema you desire
    Schema countryInfo = new Schema {
      Properties =
        new Dictionary<string, Schema> {
          {
            "title", new Schema { Type = Type.STRING, Title = "Title" }
          },
          {
            "population", new Schema { Type = Type.INTEGER, Title = "Population" }
          },
          {
            "capital", new Schema { Type = Type.STRING, Title = "Capital" }
          },
          {
            "continent", new Schema { Type = Type.STRING, Title = "Continent" }
          },
          {
            "language", new Schema { Type = Type.STRING, Title = "Language" }
          }
        },
      PropertyOrdering =
          new List<string> { "title", "population", "capital", "continent", "language" },
      Required = new List<string> { "title", "population", "capital", "continent", "language" },
      Title = "CountryInfo", Type = Type.OBJECT
    };

    var response = await client.Models.GenerateContentAsync(
        model: "gemini-2.0-flash",
        contents: "Give me information about Australia",
        config: new GenerateContentConfig {
            ResponseMimeType = "application/json",
            ResponseSchema = countryInfo
        }
    );

    string text = response.Candidates[0].Content.Parts[0].Text;
    var parsedText = JsonSerializer.Deserialize<Dictionary<string, object>>(text);
    Console.WriteLine(parsedText);
  }
}

Generate Content Stream

The usage of GenerateContentStreamAsync is similar to GenerateContentAsync, this section shows one simple example to showcase the nuance in the usage.

using System.Threading.Tasks;
using Google.GenAI;
using Google.GenAI.Types;

class GenerateContentStreamSimpleText {
  public static async Task main() {

    // assuming credentials are set up in environment variables as instructed above.
    var client = new Client();
    await foreach (var chunk in client.Models.GenerateContentStreamAsync(
        model: "gemini-2.0-flash",
        contents: "why is the sky blue?"
    )) {
        Console.WriteLine(chunk.Candidates[0].Content.Parts[0].Text);
    }
  }
}

Generate Images

using System.Threading.Tasks;
using Google.GenAI;
using Google.GenAI.Types;

public class GenerateImagesSimple {
  public static async Task main() {
    // assuming credentials are set up in environment variables as instructed above.
    var client = new Client();
    var generateImagesConfig = new GenerateImagesConfig
    {
      NumberOfImages = 1,
      AspectRatio = "1:1",
      SafetyFilterLevel = SafetyFilterLevel.BLOCK_LOW_AND_ABOVE,
      PersonGeneration = PersonGeneration.DONT_ALLOW,
      IncludeSafetyAttributes = true,
      IncludeRaiReason = true,
      OutputMimeType = "image/jpeg",
    };
    var response = await client.Models.GenerateImagesAsync(
      model: "imagen-3.0-generate-002",
      prompt: "Red skateboard",
      config: generateImagesConfig
    );
    // Do something with the generated image
    var image = response.GeneratedImages.First().Image;
  }
}

Upscale Image

Upscaling an image is only supported on the Vertex AI client.

using System.Threading.Tasks;
using Google.GenAI;
using Google.GenAI.Types;

public class UpscaleImageSimple {
  public static async Task main() {
    // assuming credentials are set up in environment variables as instructed above.
    var client = new Client();
    var upscaleImageConfig = new UpscaleImageConfig {
      OutputMimeType = "image/jpeg", EnhanceInputImage = true
    };
    var image; // Image to upscale here
    var response = await client.Models.UpscaleImageAsync(
      model: modelName, image: image, upscaleFactor: "x2",
      config: upscaleImageConfig);

    // Do something with the generated image
    var image = response.GeneratedImages.First().Image;
  }
}

Edit Image

Edit image uses a separate model from generate and upscale.

Edit image is only supported in the Vertex AI client.

using Google.GenAI;
using Google.GenAI.Types;

public class EditImageSimple {
  public static async Task main() {
    // assuming credentials are set up in environment variables as instructed above.
    var client = new Client();

    List<IReferenceImage> referenceImages = new List<IReferenceImage>();

    // Raw reference image
    var rawReferenceImage = new RawReferenceImage {
      ReferenceImage = Image.FromFile("path/to/file", "image/png"),
      ReferenceId = 1,
    };
    referenceImages.Add(rawReferenceImage);

    // Mask reference image, generated by model
    var maskReferenceImage = new MaskReferenceImage {
      ReferenceId = 2,
      Config =
          new MaskReferenceConfig {
            MaskMode = MaskReferenceMode.MASK_MODE_BACKGROUND,
            MaskDilation = 0.06,
          }
    };
    referenceImages.Add(maskReferenceImage);

    var editImageConfig = new EditImageConfig {
      EditMode = EditMode.EDIT_MODE_INPAINT_INSERTION,
      NumberOfImages = 1,
      OutputMimeType = "image/jpeg",
    };

    var editImageResponse = await vertexClient.Models.EditImageAsync(
        model: "imagen-3.0-capability-001",
        prompt: "Change the colors of [1] using the mask [2]",
        referenceImages: referenceImages,
        config: editImageConfig);

    // Do something with the generated image
    var image = editImageResponse.GeneratedImages.First().Image;
  }
}

Segment Image

Segment image is only supported in the Vertex AI client.

using Google.GenAI;
using Google.GenAI.Types;

public class SegmentImageSimple {
  public static async Task main() {
    // assuming credentials are set up in environment variables as instructed above.
    var client = new Client();

    var segmentImageConfig = new SegmentImageConfig {
      Mode = SegmentMode.BACKGROUND,
      MaxPredictions = 1,
    };

    var segmentImageResponse = await vertexClient.Models.SegmentImageAsync(
        model: modelName,
        source: new SegmentImageSource {
          Image = Image.FromFile("path/to/image.png", "image/png"),
        },
        config: segmentImageConfig);

    // Do something with the generated mask
    var mask = segmentImageResponse.GeneratedMasks.First().Mask;
  }
}