Back to agent index
Groq

Groq

Agent framework by Groq

Revolutionize your AI workflow with the Groq AI Agent, where cutting-edge LPU technology meets unparalleled speed and efficiency for next-level machine learning solutions.

CodingBusinessesClosed source
groq.com/showcases/project-agent-remix

The Groq AI Agent is a state-of-the-art solution designed to optimize AI inference by utilizing Groq's innovative Language Processing Units (LPUs). This advanced technology is specifically engineered to enhance machine learning computations, particularly for large language models (LLMs), generative AI, and other intricate AI tasks. By leveraging LPUs, the Groq AI Agent offers significant improvements in speed and efficiency, making it a valuable tool for a range of applications in the AI landscape.

Features

The Groq AI Agent boasts a variety of features tailored to support diverse AI workloads and enhance performance across different applications. Below is an overview of its key features:

Feature Description
High-Performance LPU Inference Engine Utilizes Tensor Streaming Processor (TSP) technology for efficient AI workload processing, resulting in low latency and high throughput.
Configurable Mixture of Agents (MoA) Framework Facilitates collaboration among multiple AI models to enhance accuracy and reduce costs for complex queries.
Autonomous AI Agents Enables the creation of self-learning, real-time AI agents that can operate independently.
Interactive Chat Interface Showcases the MoA architecture through a Streamlit application, allowing real-time configuration and visualization of outputs.
Scalability and Performance Offers accelerated compute performance of up to 48 PetaOPs (INT8) or 12 PFLOPs (FP16), significantly faster than existing models.
Configurability and Customization Allows users to customize agent parameters, providing flexibility for specific use cases and requirements.
Integration and Compatibility Supports integration with a Groq API key for seamless access to advanced AI capabilities.

Use cases

The Groq AI Agent can be applied across various domains, demonstrating its versatility and capability:

  • Autonomous Vehicles: Utilizing low-latency inference for real-time decision-making in complex driving environments.
  • Robotics: Enhancing robotic control systems with autonomous agents that adapt to changing conditions and tasks.
  • Advanced AI Chatbots: Leveraging the MoA framework to improve conversational AI by integrating insights from multiple models.
  • Generative AI Applications: Rapidly generating content or simulations with high performance, suitable for entertainment or educational purposes.
  • Interactive Data Analysis: Allowing users to interactively explore data with customizable AI agents that provide insights based on layered processing.

How to get started

To begin utilizing the Groq AI Agent, users can access the Groq API by signing up for a free account at console.groq.com. This will provide the necessary API key to leverage the full capabilities of the agent. Additionally, developers can explore the available documentation and resources to understand how to effectively implement and configure the agent for their specific needs.

Pricing Indication for Groq AI Models

The following are the input and output token prices per million tokens for various Groq AI models:

  • Llama 3.2 1B (Preview) 8k: $0.04 (input), $0.04 (output)
  • Llama 3.2 3B (Preview) 8k: $0.06 (input), $0.06 (output)
  • Llama 3.3 70B Versatile 128k: $0.59 (input), $0.79 (output)
  • Llama 3.1 8B Instant 128k: $0.05 (input), $0.08 (output)
  • Mixtral 8x7B Instruct 32k: $0.24 (input and output)
  • Gemma 7B 8k Instruct: $0.07 (input and output)
  • Gemma 2 9B 8k: $0.20 (input and output)
  • Llama 3 Groq 70B Tool Use Preview 8k: $0.89 (input and output)
  • Llama 3 Groq 8B Tool Use Preview 8k: $0.19 (input and output)
  • Llama Guard 3 8B 8k: $0.20 (input and output)
  • Llama 3.3 70B SpecDec 8k: $0.59 (input), $0.99 (output)

Automatic Speech Recognition (ASR) Model Pricing

  • Whisper V3 Large: $0.111 per hour transcribed
  • Whisper Large v3 Turbo: $0.04 per hour transcribed
  • Distil-Whisper: $0.02 per hour transcribed

Vision Model Pricing

  • Llama 3.2 11B Vision 8k (Preview): $0.18 per million tokens
  • Llama 3.2 90B Vision 8k (Preview): $0.90 per million tokens

Please note that the prices listed are for tokens-as-a-service and may vary depending on the specific model and deployment scenario. For detailed pricing information regarding enterprise API solutions or on-prem deployments, please contact Groq directly.