France-based startup Mistral AI has introduced its latest proprietary large language model (LLM), known as Mistral Large, into the competitive artificial intelligence market. Positioned as a competitor to established rivals, Mistral Large boasts impressive performance on three math and coding tests, outperforming several well-known LLMs, according to a post by the Paris-based company on February 26.
Despite its strong performance, Mistral Large falls short of matching products like xAI's Grok and Google's Gemini Ultra, which debuted in November and early February, respectively. While Mistral AI claims superiority over its previous models, including an AI chat interface named "Le Chat," some observers express disappointment over Mistral Large being a closed proprietary model, similar to OpenAI's latest LLM releases.
Guillaume Lample, the founder and chief scientist of Mistral AI, asserts that Mistral Large represents a significant advancement over the company's earlier offerings. With support from prominent investors such as Nvidia, Salesforce, and Andreessen Horowitz, Mistral AI secured $487 million in funding in December. The new model boasts proficiency in multiple languages, including English, French, Spanish, German, and Italian, and demonstrates an understanding of over 20,000 words in English.
While Mistral AI's first model was released under an open-source license, Mistral Large marks a shift toward closed proprietary models, akin to recent releases by OpenAI. This transition has sparked disappointment among some observers. Although third-party AI chatbot ranking platforms like Chatbot Arena have yet to assess Mistral Large, the company's earlier model, Mistral Medium, achieved a notable sixth-place ranking among over 60 LLMs on the platform.
Furthermore, Mistral AI recently announced a collaboration with Microsoft to integrate Mistral Large into Azure AI Studio and Azure Machine Learning. This partnership signals Microsoft's confidence in Mistral AI's business-focused LLM, with Mistral Large leveraging Azure's robust supercomputing infrastructure for training and scaling. According to Eric Boyd, corporate vice president of Microsoft's Azure AI Platform, the collaboration will extend to AI research and development efforts, underscoring the significance of Mistral AI's advancements in the field.
