On April 18, Meta announced the release of Llama 3, the latest version of its Large-Scale Language Model (LLM), describing it as a “major advancement over Llama 2.”
According to the company, the first two models of the current version are equipped with 8B and 70B parameters, and future models will be equipped with 400B parameters.
Mehta highlighted that Llama 3 was trained on a “large, high-quality training dataset” with more than 15 trillion tokens, seven times more than Llama 2. Llama 3 also features filtering techniques such as his NSFW filter to ensure data quality.
LLama 3 outperforms Llama 2 and competing models such as Anthropic’s Claude Sonnet, Mistral Medium, and OpenAI’s Chat GPT-3.5 more than half the time in 12 use cases.
The first release of Llama 3 is a text-based model. However, future releases are planned to be multilingual and multimodal. It also features longer contexts and demonstrates improved performance in reasoning and coding. Mehta described this as “core LLM functionality.”
The company plans to deploy Llama 3 across all major cloud providers, model API providers, and other services. The company plans to release the product “everywhere.”
Broader user access
While Llama 3 is aimed at developers, Meta also launched a new way for end users to access AI services in the US and over a dozen other countries.
The new addition is a dedicated website called Meta AI, where users can access AI-powered writing assistance, trivia games, mock job interviews, and homework help.
Meta has also integrated Meta AI with all of its products, including Facebook, Instagram, WhatsApp, and Messenger. Additionally, the company is offering this service in the US via Ray-Ban Meta smart glasses and plans to extend it to the Meta Quest VR headset as well.
CEO Mark Zuckerberg said in the announcement:
“We believe Meta AI is the most intelligent AI assistant at our disposal.”
News of Meta’s AI product expansion comes on the heels of upgrades to its competing services. ChatGPT was upgraded to GPT-4 Turbo on April 11th, and Microsoft Copilot was upgraded to GPT-4 Turbo starting in March, advancing competition among consumer AI services.
