Table of Contents
Microsoft, a leading player in the generative AI field, has diversified its AI portfolio by adding Llama 2, an open-source AI model developed by Meta Platforms, to its Azure AI Studio. Llama 2 is a powerful language model that can generate text and chat responses for various domains and tasks.
Customers can now access Llama 2 as a model-as-a-service, which means they can use it on-demand over the web without installing it on their infrastructure.
What is model-as-a-service?
Model-as-a-service (MaaS) is a cloud-based service that allows customers to use AI models such as Llama 2 via API endpoints, similar to how they would use software-as-a-service (SaaS). This reduces the complexity and cost of deploying and managing AI models, especially for enterprises that may not have the expertise or resources to do so.
In a blog post, Microsoft Corporate Vice President of Program Management for its AI Platform efforts, John Montgomery, explained:
“In Azure AI, you have been able to deploy models onto your infrastructure for a long time – simply go into the model catalog, select the model to deploy and a VM to deploy it on and you’re off to the races. But not every customer wants to think about operating infrastructure, which is why at Ignite we introduced Models as a Service, which operates models as API endpoints that you simply call, much the way you might call the Azure OpenAI Service.”
Why Llama 2?
Llama 2 is one of the most popular open-source AI models in the market, with several variants that cater to different needs and scenarios. Microsoft is making six of these variants available in Azure AI Studio in public preview, namely:
- Llama-2-7b (Text Generation)
- Llama-2-7b-Chat (Chat Completion)
- Llama-2-13b (Text Generation)
- Llama-2-13b-Chat (Chat Completion)
- Llama-2-70b (Text Generation)
- Llama-2-70b-Chat (Chat Completion)
By adding Llama 2 to its AI offerings, Microsoft is giving its customers more options and flexibility to choose the AI model that suits their use cases and budgets. Llama 2 is also a lower-cost alternative to OpenAI’s GPT-3.5 and 4 models, which Microsoft has an exclusive license to use and integrate into its products and services.
Llama 2 has been widely acclaimed as one of the best generative AI models of the year, as it has demonstrated impressive performance and versatility in various tasks and domains. Many users and enterprises prefer Llama 2 over other models because of its open-source nature and community support.
What about OpenAI?
Microsoft’s decision to include Llama 2 in its AI portfolio does not mean that it is abandoning its partnership with OpenAI, the startup that created GPT-3 and GPT-4, the largest and most advanced language models in the world. On the contrary, Microsoft is still committed to supporting and promoting OpenAI’s vision and technology.
Montgomery also announced that OpenAI’s latest AI model, GPT-4 Turbo with Vision, is now available in Azure AI Studio. This model combines natural language and computer vision capabilities, enabling it to analyze and describe images and other visual content. Customers such as Instacart and WPP are already using this model, according to Montgomery.
Moreover, Azure AI Studio provides tools for fine-tuning all the models offered, including Llama 2 and GPT-4 Turbo with Vision, to optimize them for specific tasks and domains.
What’s next for Microsoft’s AI?
Microsoft’s AI strategy is to offer a comprehensive and diverse range of AI models and services to its customers, while also investing in its research and development. The company recently released Phi-2, its small language model (SML), for research purposes only. Phi-2 is designed to be more efficient and scalable than larger models, while still achieving high-quality results.
As the AI cloud wars heat up, Microsoft is poised to remain a dominant force in the industry, with its ability to leverage its existing platforms and partnerships, as well as its innovation and vision. The question is: what other models will Microsoft add to Azure AI Studio in the future? Will it be Mistral or Deci, or something else entirely?
