simple-groq-rs

A minimal, ergonomic, async Rust client for the Groq inference API.

Groq provides lightning-fast inference for open-source models (Llama 3.1, Mixtral, Gemma, etc.) and is fully compatible with the OpenAI API format. This crate gives you a clean, lightweight way to call Groq from Rust with almost zero overhead.

Features

• Simple, idiomatic API
• Async (powered by reqwest)
• No heavy dependencies
• Easy environment variable setup
• Ready for streaming, vision, and tools (future extensions)

Installation

Add this to your Cargo.toml:

[dependencies]
simple-groq-rs = "0.1.0"

Quick Start:

• Get a free API key at console.groq.com/keys (no credit card required).
• Set the key in your environment:

export GROQ_API_KEY="gsk_your_key_here"

Code Exmaple:

use simple_groq_rs::{GroqClient, Message};

#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    let client = GroqClient::from_env()?; // Loads GROQ_API_KEY automatically

    let messages = vec![
        Message::system("You are a helpful and concise assistant."),
        Message::user("Explain Rust's borrow checker in one short paragraph."),
    ];

    let response = client
        .chat_completion("llama-3.1-70b-versatile", messages, None, None)
        .await?;

    println!("Groq response:\n{}", response);
    Ok(())
}

Popular free-tier models (as of Dec 2025):

• llama-3.1-8b-instant – fastest, great for quick tasks
• llama-3.1-70b-versatile – highest quality
• mixtral-8x7b-32768
• gemma2-9b-it

Running Examples:

This crate includes ready-to-run examples:

# Set your key (once per terminal session)
export GROQ_API_KEY="gsk_your_key_here"

# Simple chat example
cargo run --example simple

# List all models available to your account
cargo run --example list_models