Meta to drop Llama 3 400b next week — here’s why you should care

AI generated image of 3 Llamas on a chip
(Image credit: Adobe Firefly - AI generated for Future)

Meta is set to release its most powerful AI language model yet, Llama 3 400B, by the end of July 2024 and will continue to keep it open source.

First reported by The Information, this highly anticipated launch comes just months after Meta unveiled the Llama 3 family of AI models in April, which already outperformed competitors like Google's Gemma and Anthropic's Claude Sonnet in its initial 8B and 70B parameter sizes.

The Llama 3 400B model is a game-changer, boasting over 400 billion parameters and achieving near-parity with OpenAI's GPT-4 on the MMLU benchmark despite using less than half the parameters. 

This impressive feat suggests that Meta has made significant advancements in model architecture and training efficiency, potentially rivaling the performance of GPT-4 and Claude Opus — for much lower computational costs and energy consumption. 

So, how does the release of the latest LLM model from Meta affect you? Let’s talk.

What’s the latest on Llama 3 400b?

Meta AI logo on a phone

(Image credit: Shutterstock)

Meta has been hinting at the release of the 400B model since April, promising new capabilities such as multimodality, multilingual conversation, longer context windows, and stronger overall performance. While no official release date had been announced, recent sightings of a Llama 3-405B option for WhatsApp Beta users on Android 2.24.14.7 have fueled speculation about an imminent launch.

The Information reports that Meta will release its largest Llama model yet, complete with weights, on July 23. While the article doesn’t reveal its source beyond saying that it’s a company employee, Meta’s confirmation has generated significant buzz in the AI community, as researchers and developers eagerly await the opportunity to leverage this powerful tool for their projects.

Applications of Llama 3 400B are vast, ranging from advanced chatbots and virtual assistants to content generation, language translation, and beyond. With its open availability and impressive performance, this model could democratize access to cutting-edge language AI by opening it up to a wider audience at a much lower cost.

An open-source GPT-4o competitor?

The Llama 3 400B model is particularly exciting as it approaches performance parity with OpenAI's GPT-4o (Omni) model, despite using less than half the parameters. But apart from the potential benefits to cost and energy efficiency, there’s another significant advantage.

One of the most compelling aspects of Llama 3 is its open license for research and commercial use. If the 400B model is released under the same open license, it would democratize access to state-of-the-art language capabilities, allowing researchers and developers to leverage this powerful tool for their projects without relying on expensive proprietary APIs.

However, there is some uncertainty surrounding the release of the Llama 3 400B model. A recent tweet from a prominent AI leaker, "Jimmy Apples," suggests that Meta may not open-source the weights for the 400B model. While this claim should be taken with a grain of salt, as it contradicts previous statements from Meta CEO Mark Zuckerberg, it does raise questions about the future accessibility of the model.

More from Tom's Guide

Category
Arrow
Arrow
Back to MacBook Air
Brand
Arrow
Processor
Arrow
RAM
Arrow
Storage Size
Arrow
Screen Size
Arrow
Colour
Arrow
Storage Type
Arrow
Condition
Arrow
Price
Arrow
Any Price
Showing 10 of 86 deals
Filters
Arrow
Load more deals
Ritoban Mukherjee

Ritoban Mukherjee is a freelance journalist from West Bengal, India whose work on cloud storage, web hosting, and a range of other topics has been published on Tom's Guide, TechRadar, Creative Bloq, IT Pro, Gizmodo, Medium, and Mental Floss.