Meta / Llama AI Models

Open-source Llama-Guard-4-12B model from meta-llama — available for download and self-hosting on Hugging Face.

Meta Llama 3 8B Instruct — Meta's Llama open-source language model, one of the most widely deployed open models.

Open-source Llama-3.2-11B-Vision-Instruct model from meta-llama — available for download and self-hosting on Hugging Face.

This 17-billion parameter mixture-of-experts model with 128 experts, optimized for instruction-following tasks with an expansive 1 million token context window, serves developers and organizations needing cost-effective long-document processing and complex reasoning while maintaining high performance across diverse applications.

Llama 3.2 11B Vision is a multimodal model with 11 billion parameters, designed to handle tasks combining visual and textual data. It excels in tasks such as image captioning and...

Open-source Llama-Guard-3-8B model from meta-llama — available for download and self-hosting on Hugging Face.

Llama Guard 3 is a Llama-3.1-8B pretrained model, fine-tuned for content safety classification. Similar to previous versions, it can be used to classify content in both LLM inputs (prompt classification)...

Meta's latest class of model (Llama 3) launched with a variety of sizes & flavors. This 70B instruct-tuned version was optimized for high quality dialogue usecases. It has demonstrated strong...

This large language model excels at complex reasoning, multi-turn conversations, and long-form content generation with its 131K token context window, making it ideal for developers and enterprises needing cost-effective, high-performance instruction-following capabilities for production applications.

A lightweight 1-billion parameter language model optimized for efficient inference on resource-constrained devices and edge deployment, ideal for developers and organizations seeking to run conversational AI locally without requiring expensive hardware while maintaining a 131K token context window for handling longer documents and conversations.

A 17-billion parameter model with 128 experts and 262K token context window designed for developers and researchers who need efficient long-document processing, complex reasoning, and cost-effective inference across diverse language tasks.

A lightweight 3-billion-parameter language model optimized for fast inference and low resource consumption, ideal for developers and organizations seeking to deploy efficient AI applications on edge devices, mobile platforms, or cost-constrained environments while maintaining reasonable performance across general-purpose NLP tasks.

Llama Guard 4 is a Llama 4 Scout-derived multimodal pretrained model, fine-tuned for content safety classification. Similar to previous versions, it can be used to classify content in both LLM...

Meta's latest class of model (Llama 3.1) launched with a variety of sizes & flavors. This 70B instruct-tuned version is optimized for high quality dialogue usecases. It has demonstrated strong...

Meta's latest class of model (Llama 3.1) launched with a variety of sizes & flavors. This 8B instruct-tuned version is fast and efficient. It has demonstrated strong performance compared to...

Llama 3.2 3B is a 3-billion-parameter multilingual large language model, optimized for advanced natural language processing tasks like dialogue generation, reasoning, and summarization. Designed with the latest transformer architecture, it...

The Meta Llama 3.3 multilingual large language model (LLM) is a pretrained and instruction tuned generative model in 70B (text in/text out). The Llama 3.3 instruction tuned text only model...

Llama 4 Scout 17B Instruct (16E) is a mixture-of-experts (MoE) language model developed by Meta, activating 17 billion parameters out of a total of 109B. It supports native multimodal input...

Llama 4 Maverick 17B Instruct (128E) is a high-capacity multimodal language model from Meta, built on a mixture-of-experts (MoE) architecture with 128 experts and 17 billion active parameters per forward...

Llama Guard 4 is a Llama 4 Scout-derived multimodal pretrained model, fine-tuned for content safety classification. Similar to previous versions, it can be used to classify content in both LLM...