Mistral AI Partners with Microsoft and Unveils New Models
Created on February 26|Last edited on February 26
Comment
In a strategic move to expand its AI portfolio, Microsoft has entered into a partnership with French artificial intelligence startup Mistral AI, signaling a diversification of its interests in the AI domain beyond its substantial investments in OpenAI. This collaboration is set to bolster Mistral's presence in the AI market, providing essential support to bring its advanced AI models, such as Mistral Large, to a broader audience through Microsoft's Azure platform.
Mistral Large
Mistral AI has also recently unveiled Mistral Large. Mistral Large excels in complex multilingual reasoning tasks, including text understanding, transformation, and code generation. It stands out for its fluency in several languages—English, French, Spanish, German, and Italian—demonstrating an intricate understanding of grammar and cultural contexts. This model also features a 32K token context window, enabling detailed processing of long documents, and boasts precise instruction-following capabilities vital for developers when establishing moderation policies.
Benchmarks
Mistral Large showcases exceptional reasoning, knowledge handling, and multilingual capabilities, setting new benchmarks. It demonstrates superior performance in coding and math tasks, highlighting its potential utility in technical and creative domains alike.

Mistral Small
Complementing Mistral Large, Mistral AI has also introduced Mistral Small, optimized for lower latency and cost efficiency, catering to applications requiring swift response times. Mistral Small outperforms Mixtral 8x7B and has lower latency. Both models embody advanced features like JSON format mode and function calling, enhancing the interaction with structured output and enabling complex integrations with external tools.
Closed Source
It’s important to note that these new models are closed source. As Mistral AI secures a strategic partnership with a tech giant like Microsoft and places its advanced models on commercial platforms, there's a growing curiosity about whether this signifies a pivot from its open-source ethos towards a closed source model that resembles OpenAI's. The partnership indeed marks a significant step in making Mistral's AI technologies more widely available and could potentially streamline the deployment and integration of these models into various applications and services. Yet, it also raises questions about the accessibility of Mistral's technologies for research and development purposes outside commercial frameworks.
Source:
Add a comment
Tags: ML News
Iterate on AI agents and models faster. Try Weights & Biases today.