Microsoft launched Azure AI Inference SDK for .NET, streamlining access to generative AI models in the Azure AI Studio model catalog. This catalog includes models from providers like Azure OpenAI Service, Mistral, Meta, Cohere, NVIDIA, and Hugging Face, organized into three collections: Curated by Azure AI, Azure OpenAI Models, and Open Models from Hugging Face Hub. Users can request new models and deploy existing ones via managed computing solutions.
Azure AI Inference SDK for .NET. simplifies access to the models from the Azure AI Model Catalog, enabling inference tasks such as chat integration into applications.
Using the Azure AI Inference SDK for .NET involves several key steps:
- Application Setup: Creating a new .NET application within a development environment.
- SDK Installation: The Azure.AI.Inference SDK can be installed from NuGet. This can typically be achieved through the NuGet Package Manager, allowing access to the SDK’s functionalities.
- Code Implementation for Model Requests: After installation, the next step is to incorporate code that facilitates requests to the model service.
Azure AI Studio contains a model catalog that is a central hub for discovering and utilizing a wide range of models for generative AI applications. The catalog showcases hundreds of models from various providers, including Azure OpenAI Service, Mistral, Meta, Cohere, NVIDIA, and Hugging Face, as well as models developed by Microsoft.
Azure AI Studio model catalog (Source: Microsoft Blog)
The Azure AI Studio model catalog organizes its offerings into three collections for easier access to generative AI models:
- Curated by Azure AI: This collection features popular non-Microsoft open-weight and proprietary models optimized for the Azure platform, with support provided by Microsoft under the applicable service-level agreement (SLA).
- Azure OpenAI Models: These models are available exclusively through Azure OpenAI Service and include Microsoft support aligned with specific product terms and SLAs.
- Open Models from Hugging Face Hub: This collection includes hundreds of models for real-time inference, maintained by Hugging Face, with support available through their forum and support channels.
Users can also request the addition of new models via a designated form, facilitating the development and deployment of innovative AI applications. Furthermore, developers can deploy these models using managed computing solutions, including serverless APIs with flexible pay-as-you-go billing.
More details are available in the Azure AI Model Catalog documentation.