Microsoft, a major beneficiary of the generative AI surge, has diversified its AI portfolio beyond its early support for OpenAI. The tech giant has integrated OpenAI’s technology into various services like Bing and Azure. Now, Microsoft is broadening its AI offerings by incorporating Meta Platforms’ open-source AI model, Llama 2, into Azure AI Studio as a model-as-a-service.
Model-as-a-service (MaaS) parallels the concept of software-as-a-service (SaaS), offering on-demand access to AI models like Llama 2 via the web. This approach spares customers the complexity of installing these models on their own cloud servers. Microsoft’s Azure AI now allows customers to deploy models easily as API endpoints, a more user-friendly option compared to managing infrastructure.
The introduction of a range of Llama 2 models in Azure AI Studio marks a strategic move by Microsoft. It provides Azure users with more AI options, including cost-effective alternatives to OpenAI’s GPT-3.5 and 4 models. Llama 2 has gained popularity as a preferred open-source generative AI option.
Despite expanding its AI offerings, Microsoft continues to feature OpenAI’s models prominently. The recent addition of GPT-4 Turbo with Vision in Azure AI Studio exemplifies this. This model, already in use by companies like Instacart and WPP, adds the capability to analyze and describe visual content.
Microsoft’s AI strategy also includes the development of its own models. The company recently released Phi-2, a small language model, although it’s restricted to research applications and not available for commercial use, unlike Llama-2 and GPT-4 Turbo with Vision.
In the backdrop of the ongoing AI cloud wars, Microsoft’s strategy reflects a commitment to offering a diverse range of AI models to its Azure AI Studio customers, potentially including future additions like Mistral or Deci.