Vercel AI SDK for .NET

The Vercel AI SDK for .NET is a provider-agnostic C# toolkit to build AI-powered applications and agents on .NET.

It is an unofficial community port of the AI SDK. It is an independent reimplementation of the public AI SDK behavior and provider HTTP APIs. It is not a copy of Vercel’s TypeScript, not an official Vercel product, and not endorsed by Vercel. The parity target is AI SDK 7 / Language Model specification V4. See COMPATIBILITY.md for the upstream commit and the packages that are intentionally not ported.

To learn more, read the documentation.

Installation

You need a project that targets net10.0 or netstandard2.0. Run dotnet add package Vercel.AI.

Every provider ships in this one package.

Skill for Coding Agents

If you use coding agents such as Claude Code or Cursor, we highly recommend adding the Vercel AI SDK for .NET skill to your repository:

npx skills add hardkoded/vercel-ai-sdk-dotnet

Unified Provider Architecture

The SDK gives you one API to call model providers like OpenAI, Anthropic, Google, and more.

By default, string model ids go to the Vercel AI Gateway, which gives you access to all major providers. Pass a model string for any supported model:

var client = new AiClient(GatewayProvider.Create());

var result = await client.GenerateTextAsync(new GenerateTextOptions
{
    ModelId = "anthropic/claude-opus-5.5", // or "openai/gpt-6-astra", "google/gemini-3.8-flash", etc.
    Prompt = "Hello!",
});

^{snippet source | anchor}

You can also connect to a provider directly. Each provider has a Create() method that reads its API key from the same environment variable as the JavaScript provider:

var client = new AiClient(GatewayProvider.Create());
var anthropic = AnthropicProvider.Create(); // or OpenAIProvider.Create(), GoogleProvider.Create(), etc.

var result = await client.GenerateTextAsync(new GenerateTextOptions
{
    Model = anthropic.LanguageModel("claude-opus-5-5"),
    Prompt = "Hello!",
});

^{snippet source | anchor}

Usage

Generating Text

var client = new AiClient(GatewayProvider.Create());

var result = await client.GenerateTextAsync(new GenerateTextOptions
{
    ModelId = "openai/gpt-6-astra", // use Vercel AI Gateway
    Prompt = "What is an agent?",
});

Console.WriteLine(result.Text);

^{snippet source | anchor}

Generating Structured Data

var client = new AiClient(GatewayProvider.Create());

var result = await client.GenerateTextAsync(new GenerateTextOptions
{
    ModelId = "openai/gpt-6-astra",
    Output = OutputSpec.Object("""
        {
          "type": "object",
          "properties": {
            "recipe": {
              "type": "object",
              "properties": {
                "name": { "type": "string" },
                "ingredients": {
                  "type": "array",
                  "items": {
                    "type": "object",
                    "properties": {
                      "name": { "type": "string" },
                      "amount": { "type": "string" }
                    },
                    "required": ["name", "amount"],
                    "additionalProperties": false
                  }
                },
                "steps": { "type": "array", "items": { "type": "string" } }
              },
              "required": ["name", "ingredients", "steps"],
              "additionalProperties": false
            }
          },
          "required": ["recipe"],
          "additionalProperties": false
        }
        """),
    Prompt = "Generate a lasagna recipe.",
});

Console.WriteLine(result.Output);

^{snippet source | anchor}

Agents

var weatherAgent = new Agent(new AgentOptions
{
    ModelId = "openai/gpt-6-astra",
    Instructions = "You are an agent that answers questions about the weather.",
    Tools = new[]
    {
        Tool.Function(
            "weather",
            "Returns the weather in a city.",
            """{ "type": "object", "properties": { "city": { "type": "string" } }, "required": ["city"] }""",
            (arguments, cancellationToken) =>
            {
                var city = arguments.GetProperty("city").GetString();
                return Task.FromResult(JsonSerializer.Serialize(new { city, forecast = "sunny", temperature = 22 }));
            }),
    },
});

var result = await weatherAgent.GenerateAsync("What is the weather in Lisbon?");
Console.WriteLine(result.Text);

^{snippet source | anchor}

Examples

Every code block in this README comes from a project in examples. Each project is one example, and CI builds all of them.

Example	Shows
GenerateText	Generate text with a Gateway model string.
GenerateStructuredData	Generate a JSON object that matches a schema.
Agent	Run an agent that calls a tool.
Providers	Call a model through the Gateway and through a provider directly.
StreamText	Stream text as the model writes it.
Tools	Let the model call a tool and read its result.
UIMessageStream	Stream a UI message response from ASP.NET Core to a JavaScript useChat client.

Run an example with dotnet run --project examples/GenerateText. The examples call the AI Gateway when AI_GATEWAY_API_KEY is set. Without the key, they print a message and exit. Providers also calls Anthropic directly when ANTHROPIC_API_KEY is set. UIMessageStream listens on http://127.0.0.1:43123 and uses a scripted model when no key is set.

This port has no UI packages. ToUIMessageStreamResult writes the AI SDK UI message stream (text/event-stream) that a JavaScript useChat client already reads.

Tests

Run dotnet test Vercel.AI.slnx -c Release.

Unit tests mock HTTP and do not need keys. Integration tests call a live provider only when AI_GATEWAY_API_KEY or OPENAI_API_KEY is set.

Packages

The library targets net10.0 and netstandard2.0. The ASP.NET UI stream helper and the examples are net10.0 only. Versions come from MinVer. Pushing a vMAJOR.MINOR.PATCH tag packs Vercel.AI, publishes that package to nuget.org with Trusted Publishing, and the docs workflow publishes the DocFX site.

Community

Ask questions, report bugs, and share ideas in GitHub issues.

Contributing

Contributions are welcome. Read CONTRIBUTING.md before you start.

Authors

This port is maintained by Darío Kondratiuk. The AI SDK it ports is created by Vercel and Next.js team members, with contributions from the Open Source Community.

License

Apache License 2.0. Copyright 2023 Vercel, Inc. See LICENSE and NOTICE.