Meta and Microsoft expand AI partnership with the launch of Llama 2

Meta is joining forces with Microsoft to launch Meta Llama, an open source large language model developed to compete with OpenAI's GPT-4. It will be freely accessible for both research and commercial use.

Its purpose is to empower developers and organizations to create AI-powered tools and experiences. The announcement was made during Microsoft's Inspire event, highlighting Meta's support for Azure and Windows and the expanding partnership between the two tech giants.

In terms of size, Llama 2 varies from 7 to 70 billion parameters. Meta claims that it outperforms open source chat models on most benchmarks tested. For Microsoft, the introduction of Llama 2 is another attempt to keep ahead of AI competitors, such as Google Gemini. Microsoft currently utilizes OpenAI systems, including ChatGPT, in products like Microsoft Azure and Microsoft Copilot. The partnership with Meta offers more options to Microsoft's business customers, especially those interested in customizing a model to meet their specific requirements.

Alongside the launch of Llama 2, Meta has published a Responsible Use Guide. This guide is a resource for developers, outlining best practices and considerations for developing products powered by large language models (LLMs) responsibly. It covers various stages of development from inception to deployment.

With actors like OpenAI, Google, Microsoft, Meta… The race will continue with the same goal: to offer increasingly highly-trained, cutting-edge models for specific uses in fine-tuned versions, but also more lightweight models, so that they can be integrated into more accessible machines, including in the academic and corporate worlds.