Microsoft and Mistral AI forge partnership for AI advancement on Azure


Microsoft and Mistral AI have unveiled a new multi-year partnership aimed at accelerating AI innovation and introducing the Mistral Large model first on the Azure platform.

Through this collaboration, Mistral AI gains access to Azure’s advanced AI infrastructure, facilitating the rapid development and deployment of their next-generation large language models (LLMs).

Core Focus Areas of the Partnership

Supercomputing Infrastructure: Microsoft will provide Mistral AI with Azure AI supercomputing infrastructure, ensuring top-tier performance and scalability for AI training and inference workloads, particularly for Mistral AI’s flagship models.

Scale to Market: Mistral AI’s premium models will be made available to customers through the Models as a Service (MaaS) in Azure AI Studio and Azure Machine Learning model catalog. This includes Mistral Large, offering customers a diverse selection of both open-source and commercial models.

AI Research and Development: The partnership will explore collaboration on training purpose-specific models, catering to select customers’ needs, including European public sector workloads.

Expanding AI Functions with Azure and Mistral AI

In November 2023, Microsoft unveiled the integration of Mistral 7B into the Azure AI model catalog, signaling the expansion of offerings.

Mistral Large, a versatile language model proficient in various tasks including code, mathematics, and multilingual understanding, is now available on Azure AI, marking a significant step forward in AI accessibility.

Le Chat Mistral: Your Conversational Assistant

Mistral has introduced “le Chat Mistral,” a conversational assistant designed to showcase Mistral AI’s technology in a pedagogical and engaging manner.

Le Chat utilizes Mistral Large or Mistral Small models, offering users a glimpse into the capabilities of Mistral AI’s models while ensuring a non-invasive conversation experience.

Mistral Au Large: Flagship Model Unveiled

Mistral Large, the company’s flagship model, boasts top-tier reasoning capabilities and multilingual proficiency.

Available on Azure, Mistral Large empowers developers with cutting-edge text generation capabilities, making it a valuable asset for complex AI-driven applications.

Mistral Large Capacities
  • Reasoning and Knowledge: Mistral Large exhibits robust reasoning capabilities, demonstrated through performance on standard benchmarks.

  • Multi-Lingual Capacities: With native support for multiple languages, Mistral Large outperforms competitors on various benchmarks in French, German, Spanish, and Italian.

  • Math’s & Coding: Mistral Large excels in coding and math tasks, offering superior performance across popular benchmarks.

Introducing Mistral Small for Low Latency Workloads

Mistral Small, optimized for latency and cost, provides a refined solution for organizations seeking high-performance AI models without compromising speed or affordability.

Its innovative features, including JSON format mode and function calling, enhance developers’ interaction with Mistral endpoints.

Availability
  • Try Mistral Large and Mistral Small Today: Both models are available on La Plateforme and Azure, providing developers with immediate access to cutting-edge AI capabilities.
  • Azure and Mistral AI Integration: Visit the Mistral Large model card on Azure AI to begin leveraging Mistral Large’s capabilities within your Azure subscription.

Commenting on the partnership, Arthur Mensch, Chief Executive Officer, Mistral AI, said:

We are excited to embark on this partnership with Microsoft. With Azure’s cutting-edge AI infrastructure, we are achieving a new milestone in our expansion, propelling our innovative research and practical applications to new customers worldwide. Together, we are dedicated to fostering significant progress in the AI industry and providing unparalleled value to our customers and partners globally.