@elizaos/plugin-groq

Groq LLM plugin for elizaOS - Fast inference with Llama, Qwen, and other models.

This plugin provides Groq API integration for elizaOS agents, enabling ultra-fast text generation, audio transcription, and text-to-speech synthesis.

Features

• 🚀 Fast Inference - Leverage Groq's LPU for industry-leading inference speeds
• 📝 Text Generation - Generate text with Llama, Qwen, and other models
• 🎤 Audio Transcription - Transcribe audio with Whisper models
• 🔊 Text-to-Speech - Generate speech with PlayAI voices
• 🔢 Object Generation - Generate structured JSON objects
• 🎯 Tokenization - Tokenize and detokenize text

Multi-Language Support

This plugin is available for three languages:

Language	Package	Registry
TypeScript/JavaScript	@elizaos/plugin-groq	npm
Python	elizaos-plugin-groq	PyPI
Rust	elizaos-plugin-groq	crates.io

Installation

TypeScript/JavaScript (npm)

npm install @elizaos/plugin-groq
# or
bun add @elizaos/plugin-groq

Python (PyPI)

pip install elizaos-plugin-groq

Rust (crates.io)

cargo add elizaos-plugin-groq

Usage

TypeScript

import { groqPlugin } from "@elizaos/plugin-groq";

// Add to your agent's plugins
const agent = new Agent({
  plugins: [groqPlugin],
});

Python

from elizaos_plugin_groq import GroqClient, GenerateTextParams

async with GroqClient(api_key="your-api-key") as client:
    response = await client.generate_text_large(
        GenerateTextParams(prompt="What is the nature of reality?")
    )
    print(response)

Rust

use elizaos_plugin_groq::{GroqClient, GenerateTextParams};

#[tokio::main]
async fn main() -> anyhow::Result<()> {
    let client = GroqClient::new("your-api-key", None)?;

    let response = client.generate_text_large(GenerateTextParams {
        prompt: "What is the nature of reality?".to_string(),
        ..Default::default()
    }).await?;

    println!("{}", response);
    Ok(())
}

Configuration

Set the following environment variables:

Variable	Required	Default	Description
GROQ_API_KEY	Yes	-	Your Groq API key
GROQ_BASE_URL	No	https://api.groq.com/openai/v1	Custom API base URL
GROQ_SMALL_MODEL	No	openai/gpt-oss-20b	Model for small tasks
GROQ_LARGE_MODEL	No	llama-3.3-70b-versatile	Model for large tasks
GROQ_TTS_MODEL	No	canopylabs/orpheus-v1-english	Text-to-speech model
GROQ_TTS_VOICE	No	troy	TTS voice name
GROQ_TTS_RESPONSE_FORMAT	No	wav	TTS response format

Model Capabilities

This plugin provides handlers for the following elizaOS model types:

Model Type	Description
TEXT_SMALL	Fast text generation with smaller models
TEXT_LARGE	High-quality text generation with larger models
OBJECT_SMALL	JSON object generation (small)
OBJECT_LARGE	JSON object generation (large)
TRANSCRIPTION	Audio transcription with Whisper
TEXT_TO_SPEECH	Speech synthesis with PlayAI
TEXT_TOKENIZER_ENCODE	Tokenize text to tokens
TEXT_TOKENIZER_DECODE	Detokenize tokens to text

Development

Building from Source

# TypeScript
bun install
bun run build

# Rust
cd rust && cargo build --release

# Python
cd python && pip install -e ".[dev]"

Running Tests

# TypeScript
bun run test

# Rust
cd rust && cargo test

# Python
cd python && pytest

Linting

# TypeScript
bun run format:check

# Rust
cd rust && cargo clippy

# Python
cd python && ruff check .

License

MIT License - see LICENSE for details.

Links

• elizaOS Documentation
• Groq Console
• Groq API Documentation