Tanukie is a unified LLM client with:

• A clean, consistent API across providers (OpenAI, Groq, and more)
• Built-in retry logic with exponential backoff
• Automatic model detection from catalog
• Both sync and async interfaces
• Ergonomic macros for messages and options

Installation

Add to your Cargo.toml:

[dependencies]
tanukie = "0.1"
tokio = { version = "1", features = ["full"] }

Quick Start

use tanukie::{Client, messages};

#[tokio::main]
async fn main() -> tanukie::Result<()> {
    let client = Client::new();

    let msgs = messages![
        system: "You are a helpful assistant.",
        user: "What is the capital of France?",
    ];

    let response = client.agenerate("gpt-4o-mini", msgs).await?;
    println!("{}", response.text);

    Ok(())
}

With Options

use tanukie::{Client, messages, options};

let response = client.agenerate_with(
    "llama-3.1-8b-instant",  // Auto-detects Groq
    messages![user: "Write a haiku about Rust."],
    options![
        temperature: 0.9,
        max_tokens: 100u32,
    ],
).await?;

Sync API

// Blocking/sync version
let response = client.generate("gpt-4o-mini", msgs)?;

Supported Providers

Environment Variables

Set your API keys:

export OPENAI_API_KEY="sk-..."
export GROQ_API_KEY="gsk_..."

Contributing

Can't find your favorite model or provider? Open an issue and we'll add it!