Mistral established an important strategic partnership with Microsoft and announced the Mistral Large artificial intelligence model, which it describes as its newest and largest model for businesses. The company claims that this larger model performs well against models like Gemini and GPT.
What can we expect from Mistral Large?
Available starting today, Mistral Large is designed as a text production model that can handle complex multilingual reasoning tasks, including text comprehension, transformation and code generation.
According to the MMLU benchmark results shared by the company, this model performs to be the second best overall model available via API after GPT-4.
Mistral states that the large model will be primarily available via API, but will also be available via Azure AI thanks to the new partnership with Microsoft. The company has also released a chat app regarding its new partnership to help businesses get an idea of what it has to offer.
As a multilingual model, Mistral Large can understand, reason and produce text with native fluency not only in English but also in other languages such as French, Spanish, German and Italian.
He states that other companies such as Google and OpenAI also offer multilingual models, but that Mistral will achieve better results by emphasizing that it has the ability to understand the grammatical and cultural context in a nuanced way for all languages.
The model has a context window of 32K tokens. It can process large documents and remember information. It also has the ability to follow precise instructions, allowing developers to design their own moderation policies and native function calls.
We’ll see how the Mistral Large performs live, especially when compared to larger models like the Gemini 1.5, which offers support for up to 1 million tokens. Mistral claims that the new larger model does a pretty good job against rival models. What are you thinking? Please don’t forget to share your thoughts with us in the comments section.