Azure AI Foundry Models Find the right model from exploration to deployment all in one place. Accelerate innovation with models from Microsoft, OpenAI, DeepSeek, Mistral, xAI, Cohere, and Meta. Get started with Azure Create with Azure AI Foundry Overview Overview Models Security Benefits Customer stories News and resources FAQ OVERVIEW Smarter model selection starts here Find the right model for every use case Innovate faster using more than 11,000+ models packed for out-of-the-box-use and shared computer resources. Learn more Deploy models where you need them Easily integrate AI models into your applications without having to provision or manage infrastructure. Learn more Optimize model selection Analyze model metrics with standard datasets. Deploy model router to optimize costs and performance at runtime. Learn more Swap and compare models easily Easily switch models and compare performance with the Azure AI model inference API. Learn more MODELS Choose from more than 11,000+ models Foundry Models offer a rich and diverse collection of models designed to meet every AI need. Browse the catalog Models direct from Microsoft Models from partners and community Previous Next OpenAI Foundation models that exceed benchmark performance across image, video, and text. Learn more Meta Pre-trained, open language models ranging from 7 billion to 70 billion parameters. Learn more Mistral AI Accelerate AI innovation and achieve state-of-the-art reasoning performance. Learn more DeepSeek DeepSeek is a Chinese company that trains models at a significantly lower cost. DeepSeek R1 is now available on Azure AI Foundry and GitHub. Learn more xAI Supercharge enterprise AI with deep reasoning, domain expertise, and blazing-fast scalability with Grok. Learn more Black Forest Labs Harness the power of industry-leading image generation capabilities with the Flux family of models. Learn more Nixtla Pre-trained, generative AI transformer models for time-series analysis. Learn more AI21 Foundation chat completion models for enterprise that accelerate the use of generative AI for production. Learn more NTT Data A high-performance, lightweight Japanese and English SLM with fine-tuning for secure hybrid deployment. Learn more Core42, a G42 company Leading Arabic language model JAIS accelerates the growth of a vibrant Arabic language AI ecosystem. Learn more NVIDIA NIM Microservices NVIDIA NIM is a set of easy-to-use microservices designed to accelerate the deployment of generative AI across enterprises. Learn more Stability AI Deliver exceptional text-to-image generation with superior quality and prompt adherence. Learn more Phi Small language models for building generative AI applications with better latency and lower costs. Learn more Cohere A leading large language model for retrieval-augmented generation capabilities. Learn more Hugging Face Thousands of models spanning categories from text generation to image analysis. Learn more Back to tabs Security Embedded security and compliance 34,000 Full-time equivalent engineers dedicated to security initiatives at Microsoft. Learn more 15,000 Partners with specialized security expertise. Learn more >100 Compliance certifications, including over 50 specific to global regions and countries. Learn more Learn more about security on Azure BENEFITS Accelerate AI innovation Previous Slide Next Slide Task-centric model discovery Explore AI models by task and use the playground to experiment with sample queries. Learn more Ready-to-use fine-tuning Accelerate AI projects with ready-to-use fine-tuning pipelines—no setup needed. Learn more Evaluate using your own data Assess model performance using your own datasets, compare metrics, and measure improvements. Learn more Effortless AI deployment Experience hassle-free managed instances with automatic scaling, seamless traffic management, and secure hosting. Learn more CUSTOMER STORIES Customers are innovating with Azure AI Foundry Models View all Azure AI stories Previous Slide Next Slide “We’re using Azure to build more and better AI models for RapidRead to help support the high demand on veterinary radiologists. Additionally, the benefit of scale helps us boost the accuracy of our AI models and expand our operations.” Jerry Martin, VP of Research & Development, Mars Read the story “When it came to AI models, the Azure AI platform was miles ahead for enterprise-segment offerings compared to alternatives.” Lenin Gali, Chief Digital and Business Officer, Atomicwork Read the story NEWS AND RESOURCES Learn about Foundry Models Previous Slide Next Slide News DeepSeeK R1 DeepSeek R1 is now available on Azure AI Foundry and GitHub. Read the blog Blog DeepSeek R1 Pricing DeepSeek R1: Improved Performance, Higher Limits, and Transparent Pricing. Read the blog News NVIDIA NIM Microservices Accelerating agentic workflows with Azure AI Foundry. Read the blog News Meta Llama 3 Microsoft introduces Meta Llama 3 models to Azure AI Foundry. Read the blog Announcement Cohere Command R+ Microsoft adds Command R+ from Cohere to Foundry Models. Read the blog Blog Hosted fine-tuning using MaaS Build language model apps using model as a service (MaaS) by offering access to Llama 2 as an API. Read the blog Announcement Mistral Large Microsoft adds Mistral Large to Foundry Models. Read the blog Research Improve model training Combine retrieval-augmented generation and fine-tuning for better domain adaptation. Learn more Research Build safer AI apps GoEx helps developers create safer and more private Agentic AI applications. Learn more FAQ Frequently asked questions Expand all Collapse all 01/ What are Azure AI Foundry Models? Foundry Models are a hub for discovering foundation models. The catalog includes some of the most popular large language and vision foundation models curated by Microsoft, OpenAI, DeepSeek, xAI, Hugging Face, Meta, Mistral AI, Cohere, Deci, Stability AI, Nixtla, and NVIDIA. These models are packaged for out-of-the-box use and are optimized for use in Azure AI Foundry. 02/ What is models as a service (MaaS)? MaaS or Serverless API is a deployment type that allows developers to access and use a variety of models hosted on Azure without having to provision GPUs or manage back-end operations. MaaS offers inference APIs and hosted fine-tuning for models such as Meta Llama2, Meta Llama 3, Mistral Large, and others. 03/ How much does MaaS/Serverless API cost? MaaS charges are based on the number of tokens used for inference and the amount of data used for fine-tuning. The pricing varies depending on the model and the region. To see the price for individual models, search for the model in . 04/ How can I start using MaaS? MaaS is available in . 05/ How many models are available today through MaaS? Today, we have Meta’s Llama 2 and Llama 3 offerings, Mistral AI’s Mistral Large, Cohere models, Core42 JAIS, Nixtla models Stability and Deepseek models. Next steps Choose the Azure account that’s right for you Pay as you go or try Azure free for up to 30 days. Get started with Azure Azure solutions Azure cloud solutions Solve your business problems with proven combinations of Azure cloud services, as well as sample architectures and documentation. Explore Azure AI solutions Business Solution Hub Find the right Microsoft Cloud solution Browse the Microsoft Business Solutions Hub to find the products and solutions that can help your organization reach its goals. Explore Microsoft solutions , Azure AI Foundry Models Find the right model from exploration to deployment all in one place. Accelerate innovation with models from Microsoft, OpenAI, DeepSeek, Mistral, xAI, Cohere, and Meta., An efficient AI solution for diverse text and image tasks, including text to image, image to image, inpainting, and prompt transformation..