AISDK

An open-source, provider-agnostic Rust library for building AI-powered applications, inspired by the Vercel AI SDK. It provides a type-safe interface for interacting with Large Language Models (LLMs) and offers seamless support for Rust backend frameworks as well as popular UI frameworks like React, Solid, Vue, Svelte, and more.

To learn more about how to use the AI SDK, check out our Documentation and API Reference.

Installation

cargo add aisdk

Usage

Enable Providers of your choice such as OpenAI, Anthropic, Google, and more

Example with OpenAI provider

cargo add aisdk --features openai

Basic Text Generation

use aisdk::core::LanguageModelRequest;
use aisdk::providers::openai::OpenAI;

#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {

    let openai = OpenAI::gpt_5();

    let result = LanguageModelRequest::builder()
        .model(openai)
        .prompt("What is the meaning of life?")
        .build()
        .generate_text() // or stream_text() for streaming
        .await?;

    println!("Response: {:?}", result.text());
    Ok(())
}

Agents

Defining a Tool

Use the #[tool] macro to expose a Rust function as a callable tool.

use aisdk::core::Tool;
use aisdk::macros::tool;

#[tool]
/// Get the weather information given a location
pub fn get_weather(location: String) -> Tool {
    let weather = match location.as_str() {
        "New York" => 75,
        "Tokyo" => 80,
        _ => 70,
    };
    Ok(weather.to_string())
}

Using Tools in an Agent

Register tools with an agent so the model can call them during its reasoning loop.

use aisdk::core::{LanguageModelRequest, utils::step_count_is};
use aisdk::providers::openai::OpenAI;

#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {

    let result = LanguageModelRequest::builder()
        .model(OpenAI::gpt_4o())
        .system("You are a helpful assistant.")
        .prompt("What is the weather in New York?")
        .with_tool(get_weather())
        .stop_when(step_count_is(3)) // Limit agent loop to 3 steps
        .build()
        .generate_text()
        .await?;

    println!("Response: {:?}", result.text());
    Ok(())
}

Structured Output

Define your target output format.

use serde::Deserialize;
use schemars::JsonSchema;

#[derive(JsonSchema, Deserialize, Debug)]
struct User {
    name: String,
    age: u32,
    email: Option<String>,
}

Use the schema attribute to infer the structure of the output.

use aisdk::core::LanguageModelRequest;
use aisdk::providers::openai::OpenAI;

#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {

    let user: User = LanguageModelRequest::builder()
        .model(OpenAI::gpt_5())
        .prompt("Generate a random user")
        .schema::<User>()
        .build()
        .generate_text()
        .await?
        .into_schema()?;

    println!("Name: {}", user.name);
    println!("Age: {}", user.age);
    println!("Email: {}", user.email.unwrap_or_default());
    Ok(())
}

Prompts

The AISDK prompt feature provides a powerful, file-based template system for managing AI prompts using the Tera template engine. It allows you to create reusable prompt templates with variable substitution, conditionals, loops, and template inclusion. See Examples for more template examples. Enable with cargo add aisdk --features prompt

Roadmap

• Agents
• Tool Execution
• Prompt Templating
• Structured Output (JSON Schema)
• Language Model Request Support (Text Generation, Streaming)
• Compatible with Vercel AI SDK UI (React, Solid, Vue, Svelte, …)
• Embedding Model Request Support
• Image Model Request Support
• Voice Model Request Support
• Additional Providers

Contributing

We welcome contributions! Please see CONTRIBUTING.md for guidelines.

License

Licensed under the MIT License. See LICENSE for details.