Meta Announces Llama 2, An Open-Source AI Model Rivaling Proprietary Alternatives
Meta (formerly known as Facebook) has revealed its plans to launch a commercial version of its open-source artificial intelligence model, Llama. The new model, aptly named Llama 2, is poised to provide startups and businesses with a cost-effective alternative to expensive proprietary AI models offered by tech giants like OpenAI and Google.
In an official blog post, Meta stated that Llama 2 would be distributed by Microsoft through its Azure cloud service and would be compatible with the Windows operating system. Microsoft was referred to as "our preferred partner" for the model's release.
While Llama was previously accessible only to select academics for research purposes, the upgraded Llama 2 will also be available for direct download and through Amazon Web Services, Hugging Face, and other providers, according to announcements made by Meta CEO Mark Zuckerberg.
Zuckerberg emphasised the significance of open-source technology, stating that it fosters innovation by enabling more developers to utilise cutting-edge tools. He believes that the broader availability of advanced models like Llama 2 could potentially challenge the established dominance of closed-source models, such as those offered by Microsoft-backed OpenAI via Azure.
Comparisons between Llama and models powering OpenAI's ChatGPT and Google's Bard chatbot have shown that Llama already holds its ground. However, Llama 2 boasts a substantial improvement, having been trained on 40 per cent more data than its predecessor, with over 1 million human annotations used to refine the quality of its outputs.
The announcement follows a trend among major cloud providers, including Google and Amazon, who have been keen on offering various AI models to business customers. Amazon, for instance, promotes access to Claude AI from the notable startup Anthropic alongside its own family of Titan models. Similarly, Google plans to make Claude and other models available to its cloud customers.
