Meta has claimed that its new AI model is the first open source system that will rival products from competitors such as OpenAI and Anthropic.
In a blog post, the company said its new model, with the unwieldy name Llama 3.1 405B, “is competitive” with others, including those from OpenAI and Anthropic, “on a variety of tasks.”
If true, it would mean that for the first time, one of the world’s most powerful AI models would be available without a middleman charging for access or controlling what its technology is used for.
“Developers can fully customize models for their needs and applications, train them on new datasets, and make additional adjustments,” Meta said. “This allows the broader developer community and the entire world to more fully understand the power of generative AI. Developers can fully customize their applications and run them in any environment… all without sharing data with Meta.”
Those who use Llama in Meta’s own apps (where it is available, for now, only in the US) will have additional layers of “security,” the company says.
These are also open source and the company has no way to force others to apply them to their own uses of the model.
“I believe open source is necessary for a positive future for AI,” Meta co-founder Mark Zuckerberg wrote in a blog post. “AI has more potential than any other modern technology to increase human productivity, creativity, and quality of life, and to accelerate economic growth while driving progress in medical and scientific research. Open source will ensure that more people around the world have access to the benefits and opportunities of AI, that power is not concentrated in the hands of a small number of companies, and that the technology can be deployed more uniformly and safely across society.”
Zuckerberg admitted that “bad actors may be able to use the intelligence of AI models to create entirely new harms,” but maintained that “I think it will be better to live in a world where AI is widely deployed so that the larger actors can check the power of the smaller bad actors.”
So far, Meta has been marking its own task in the matter of its model’s power. The model’s sheer size puts it on par with larger competing systems. But until third parties can conduct fair testing between the Llama 3.1 405B and its peers like GPT-4o, there’s no guarantee that sheer size will match the effectiveness of the current leaders in the field.
The AI model is currently only available to regular users in 22 countries, including the US, via Meta.ai. Meta reportedly avoided an EU launch due to concerns about the bloc’s regulatory framework around AI and data protection, but the full open-access Llama 3.1 model is available globally for those who can run it.