All You Need to Know: Meta’s Llama, an open generative AI model

Meta is stepping up its game in the artificial intelligence (AI) competition with the launch of a new version of Meta AI along with its new open-source AI model, Llama 3.

Now accessible on all Meta platforms, this virtual assistant is powered by Llama 3.

Here’s all you need to know about Meta’s newest large language model (LLM) and AI assistant.

What is Llama?

Llama 3.1 8B, Llama 3.1 70B, and Llama 3.1 405B are the most recent versions, having been made available in July 2024.

They receive training on multiple language web pages, publicly available web files and code, and artificial intelligence (i.e. data produced by additional AI models).

Small and portable, the Llama 3.1 8B and Llama 3.1 70B are designed to operate on a variety of devices, including servers and laptops.

On the other hand, Llama 3.1 405B is a large-scale model that needs data center hardware (barring any changes).

While Llama 3.1 405B is faster, Llama 3.1 8B and Llama 3.1 70B are less capable. Indeed, they are “distilled” versions of 405B, with low latency and storage overhead optimized.

What can Llama do?

Llama is capable of coding, answering simple math questions, and summarizing documents in eight different languages (English, German, French, Italian, Portuguese, Hindi, Spanish, and Thai), just like other generative AI models.

None of the Llama models can currently process or generate images, but that may change in the near future. Instead, it is limited to text-based workloads, such as spreadsheet and PDF analysis.

It is possible to set up the most recent Llama models to use external tools, applications, and APIs to accomplish tasks. They can use Brave Search to find answers to current events queries, the Wolfram Alpha API for queries pertaining to math and science, and a Python interpreter for code validation right out of the box.

Furthermore, Meta states that the Llama 3.1 models are capable of using some tools that they have never seen before (though it is debatable if they can use those tools consistently).

Where can I use Llama?

In case you want to have a casual conversation, Llama powers the Meta AI chatbot experience on various platforms such as Facebook Messenger, WhatsApp, Instagram, Oculus, and Meta.ai.

Across the majority of well-known cloud platforms, developers using Llama can download, utilize, and optimize the model.

According to Meta, more than 25 partners—including Nvidia, Databricks, Groq, Dell, and Snowflake—are hosting Llama.