Gonka

Modern .NET SDK for direct Gonka Network inference with OpenAI-compatible chat completions and ECDSA request signing.

Get started Get support

Manual OpenAPI spec

Built from a maintained OpenAPI description because Gonka does not currently publish a direct SDK OpenAPI document.

Direct network auth

Signs each request with secp256k1 ECDSA and sends the Gonka requester headers required by provider endpoints.

Modern .NET

Targets current .NET practices including nullability, trimming, NativeAOT awareness, and source-generated serialization.

Docs from examples

Examples stay in sync between the README, MkDocs site, and integration tests through the AutoSDK docs pipeline.

Usage

using Gonka;

using var client = await GonkaClient.CreateFromEnvironmentAsync();

var response = await client.CreateChatCompletionAsync(
    new CreateChatCompletionRequest
    {
        Model = "Qwen/Qwen3-235B-A22B-Instruct-2507-FP8",
        Messages =
        [
            new ChatCompletionMessage
            {
                Role = ChatCompletionMessageRole.User,
                Content = "Hello, Gonka!",
            },
        ],
    });

Set GONKA_PRIVATE_KEY plus either GONKA_ENDPOINTS (https://host/v1;gonka1provider...) or GONKA_SOURCE_URL. GONKA_ADDRESS can override the derived requester address.

Generate

Basic example showing how to create a client and make a request.

// Direct Gonka requests are signed with your Gonka private key.
// Provide either GONKA_ENDPOINTS (`https://host/v1;gonka1provider...`) or GONKA_SOURCE_URL for endpoint discovery.
using var client = await GetAuthenticatedClientAsync().ConfigureAwait(false);

var model = Environment.GetEnvironmentVariable("GONKA_CHAT_MODEL") is { Length: > 0 } modelValue
    ? modelValue
    : "Qwen/Qwen3-235B-A22B-Instruct-2507-FP8";

var response = await client.CreateChatCompletionAsync(
    new CreateChatCompletionRequest
    {
        Model = model,
        Messages =
        [
            new ChatCompletionMessage
            {
                Role = ChatCompletionMessageRole.User,
                Content = "Hello, Gonka!",
            },
        ],
    }).ConfigureAwait(false);

Stream Chat

Streaming example showing how to read server-sent chat completion chunks.

// Streaming requests use the same direct Gonka signing layer as non-streaming requests.
using var client = await GetAuthenticatedClientAsync().ConfigureAwait(false);

var model = Environment.GetEnvironmentVariable("GONKA_CHAT_MODEL") is { Length: > 0 } modelValue
    ? modelValue
    : "Qwen/Qwen3-235B-A22B-Instruct-2507-FP8";

var chunks = new List<CreateChatCompletionResponse>();
await foreach (var chunk in client.CreateChatCompletionStreamingAsync(
    new CreateChatCompletionRequest
    {
        Model = model,
        Messages =
        [
            new ChatCompletionMessage
            {
                Role = ChatCompletionMessageRole.User,
                Content = "Write one short sentence about Gonka.",
            },
        ],
    }))
{
    chunks.Add(chunk);
}

Microsoft.Extensions.AI Chat Client

MEAI example showing how to use Gonka through the standard IChatClient abstraction.

// The Gonka client implements Microsoft.Extensions.AI.IChatClient for shared chat workflows.
using var client = await GetAuthenticatedClientAsync().ConfigureAwait(false);
Microsoft.Extensions.AI.IChatClient chatClient = client;

var model = Environment.GetEnvironmentVariable("GONKA_CHAT_MODEL") is { Length: > 0 } modelValue
    ? modelValue
    : "Qwen/Qwen3-235B-A22B-Instruct-2507-FP8";

var response = await chatClient.GetResponseAsync(
    [new Microsoft.Extensions.AI.ChatMessage(Microsoft.Extensions.AI.ChatRole.User, "Write one short sentence about Gonka.")],
    new Microsoft.Extensions.AI.ChatOptions { ModelId = model }).ConfigureAwait(false);

Support

Bugs

Open an issue in tryAGI/Gonka.

Ideas and questions

Use GitHub Discussions for design questions and usage help.

Community

Join the tryAGI Discord for broader discussion across SDKs.

Acknowledgments

This project is supported by JetBrains through the Open Source Support Program.