Llama 4 Scout

Llama 4 Scout Description

Overview

Llama 4 Scout is a lightweight, multimodal AI model capable of processing text and images. Its architecture allows it to run efficiently on a single Nvidia H100 GPU while handling a context window of up to 1 million tokens.

Key Features

• Extensive Context Window: Supports up to 1 million tokens, facilitating tasks like multi-document summarization and long-form code reasoning.​
• Efficient Performance: Operates on a single Nvidia H100 GPU, optimizing resource utilization.​
• Multimodal Capabilities: Processes both text and images, enhancing versatility.​

Intended Use

• Multi-Document Summarization: Efficiently summarizes information from multiple documents.​
• Code Analysis: Assists in understanding and reasoning over extensive codebases.​
• Content Parsing: Processes large volumes of text and images for various applications.​

Technical Details

Architecture

The model employs Meta’s mixture-of-experts (MoE) framework with active parameter counts of 109 billion. Scout uses 16 experts for task-specific activation.

Training Data

Trained on curated datasets including multilingual corpora, image datasets, and synthetic reasoning examples.

Usage

Code Samples

‍

API Documentation:

Detailed API Documentation is available here.

Ethical Guidelines

Llama 4 Scout has implemented safeguards against misuse such as generating harmful content or violating user privacy during tool integrations.

Licensing

Custom Llama 4 Community License