Llama 3.1

!logo

​Meta Llama 3.1 is a collection of multilingual large language models (LLMs) available in 8B, 70B and 405B parameter sizes. These models are designed for text-based tasks, including chat and content generation. The instruction-tuned versions available here are optimized for multilingual dialogue use cases and have demonstrated superior performance compared to many open-source and commercial chat models on common industry benchmarks.

Intended uses

• Assistant-like chat: Instruction-tuned text-only models are optimized for multilingual dialogue, making them ideal for developing conversational AI assistants. ​

• Natural language generation tasks: Pretrained models can be adapted for various text-based applications, such as content creation, summarization, and translation. ​

• Synthetic data generation: Utilize the outputs of Llama 3.1 to create synthetic datasets, which can aid in training and improving other models. ​

• Model distillation: Leverage Llama 3.1 to enhance smaller models by transferring knowledge, resulting in more efficient and specialized AI systems, or by using it as a base model to fine-tune based on the knowledge of other bigger models (see deepseek-r1-distill-llama as an example) ​

• Research purposes: Employ Llama 3.1 in academic and scientific research to explore advancements in natural language processing and artificial intelligence.

Characteristics

Available model variants

¹: VRAM estimated based on model characteristics.

latest → 8B-Q4_K_M

Use this AI model with Docker Model Runner

First, pull the model:

bash
docker model pull ai/llama3.1

Then run the model:

bash
docker model run ai/llama3.1

For more information on Docker Model Runner, explore the documentation.

Benchmark performance

Links

• [***]
• The Llama 3 Herd of Models