MENLO PARK — Meta and Microsoft have announced the launch of Llama 2, the latest iteration of Meta’s open-source large language model, now available for both research and commercial use. This development marks an expansion of the long-standing partnership between the two tech giants, with Microsoft becoming the preferred partner for Llama 2. The model is available through platforms such as Microsoft Azure, AWS, and Hugging Face, enabling developers and businesses to integrate it seamlessly into their operations.

Llama 2 introduces a robust set of tools, including model weights and pretrained versions for conversational AI. It empowers a broad range of users, from startups to researchers, to experiment and innovate in generative AI without the heavy resource burden traditionally required for such advanced models. This open approach encourages collaboration across the tech, academic, and policy communities to ensure the model’s safe and responsible development.

Meta emphasizes that open access to Llama 2 will promote the responsible use of AI by allowing developers to rigorously test and address potential vulnerabilities. To support this effort, Meta has implemented various safety measures, including red-teaming exercises and transparency reports, as well as guidelines for responsible usage.

Through this partnership, Meta and Microsoft aim to democratize access to advanced AI technologies, empowering innovators globally while promoting responsible AI practices. Meta also announced initiatives like the Open Innovation AI Research Community and the Llama Impact Challenge to further engage the public in exploring the positive applications of Llama 2 in areas like education and the environment.