What is Azure AI model inference service?
Azure AI models inference service provides access to the most powerful models available in the Azure AI model catalog. Coming from the key model providers in the industry including OpenAI, Microsoft, Meta, Mistral, Cohere, G42, and AI21 Labs; these models can be integrated with software solutions to deliver a wide range of tasks including content generation, summarization, image understanding, semantic search, and code generation.
The Azure AI model inference service provides a way to consume models as APIs without hosting them on your infrastructure. Models are hosted in a Microsoft-managed infrastructure, which enables API-based access to the model provider's model. API-based access can dramatically reduce the cost of accessing a model and simplify the provisioning experience.
Models
You can get access to the key model providers in the industry including OpenAI, Microsoft, Meta, Mistral, Cohere, G42, and AI21 Labs. Model providers define the license terms and set the price for use of their models. The following list shows all the models available:
| Model provider | Models |
|---|---|
| AI21 Labs | - AI21-Jamba-1.5-Mini - AI21-Jamba-1.5-Large |
| Azure OpenAI | - o1-preview (Request Access) - o1-mini (Request Access) - gpt-4o-mini - gpt-4o - text-embedding-3-small - text-embedding-3-large |
| Cohere | - Cohere-command-r-plus-08-2024 - Cohere-command-r-08-2024 - Cohere-embed-v3-multilingual - Cohere-embed-v3-english - Cohere-command-r-plus - Cohere-command-r |
| Meta AI | - Meta-Llama-3-8B-Instruct - Meta-Llama-3-70B-Instruct - Meta-Llama-3.1-8B-Instruct - Meta-Llama-3.1-70B-Instruct - Meta-Llama-3.1-405B-Instruct - Llama-3.2-11B-Vision-Instruct - Llama-3.2-90B-Vision-Instruct |
| Mistral AI | - Mistral-Small - Mistral-Nemo - Mistral-large - Mistral-large-2407 |
| Microsoft | - Phi-3-mini-4k-instruct - Phi-3-medium-4k-instruct - Phi-3-mini-128k-instruct - Phi-3-medium-128k-instruct - Phi-3-small-8k-instruct - Phi-3-small-128k-instruct - Phi-3.5-vision-instruct - Phi-3.5-mini-instruct - Phi-3.5-MoE-instruct |
You can decide and configure which models are available for inference in the created resource. When a given model is configured, you can then generate predictions from it by indicating its model name or deployment name on your requests. No further changes are required in your code to use it.
To learn how to add models to the Azure AI model inference resource and use them read Add and configure models to Azure AI Models in Azure AI model inference.
Pricing
Models that are offered by non-Microsoft providers (for example, Meta AI and Mistral models) are billed through the Azure Marketplace. For such models, you're required to subscribe to the particular model offering in accordance with the Microsoft Commercial Marketplace Terms of Use. Users accept license terms for use of the models. Pricing information for consumption is provided during deployment.
Models that are offered by Microsoft (for example, Phi-3 models and Azure OpenAI models) don't have this requirement, and they are billed via Azure meters as First Party Consumption Services. As described in the Product Terms, you purchase First Party Consumption Services by using Azure meters, but they aren't subject to Azure service terms.