Many of the tech giants want to sell artificial intelligence to the masses, but Mark Zuckerberg is giving away what Meta considers to be one of the best AI models in the world for free.
Meta on Monday released Llama, the largest and most capable version of its large-scale language model, for free. Meta hasn’t disclosed how much it cost to develop Llama 3.1, but Zuckerberg recently told investors the company is spending billions of dollars on AI development.
With this latest release, Meta is showing that the closed approach favored by most AI companies isn’t the only way to develop AI. But the company also puts itself at the center of a debate about the dangers of releasing AI without controls. By default, Meta trains Llama to ensure that the model doesn’t produce harmful outputs, but the model can be modified to remove such safeguards.
Meta says Llama 3.1 is as smart and useful as the best commercial offerings from companies like OpenAI, Google, and Anthropic. In certain benchmarks that measure AI progress, Meta says the model is the smartest AI on the planet.
“It’s very exciting,” said Percy Liang, an associate professor at Stanford University who tracks open source AI. Liang said that if developers feel the new models perform on par with industry-leading models such as OpenAI’s GPT-4o, many of them could migrate to Meta’s products. “It’ll be interesting to see how usage evolves,” he said.
In an open letter posted to coincide with the new model’s release, Meta CEO Zuckerberg compared Llama to the open-source Linux operating system. When Linux became popular in the late ’90s and early 2000s, many large tech companies invested in closed alternatives and criticized open-source software as risky and unreliable. Today, however, Linux is widely used in cloud computing and serves as the core of the Android mobile OS.
“I believe AI will evolve in a similar way,” Zuckerberg wrote in the letter. “Currently, a few tech companies are developing state-of-the-art closed models, but open source is rapidly closing the gap.”
But Meta’s decision to offer its AI for free isn’t for selfish reasons. Previous Llama releases have helped the company secure an influential position among AI researchers, developers and startups. Liang also points out that Meta imposes restrictions on the use of Llama 3.1, for example limiting the scale at which models can be used in commercial products, making it not truly open source.
The new version of Llama has 405 billion parameters or tunable elements. Meta has already released two smaller versions of Llama 3, one with 70 billion parameters and another with 8 billion parameters. Meta has also released upgraded versions of these models today, branded Llama 3.1.
Llama 3.1 is too large to run on regular computers, but Meta says a number of cloud providers, including Databricks, Groq, AWS and Google Cloud, will offer hosting options to allow developers to run custom versions of the model. The model will also be accessible on Meta.ai.
