France-based start-up, Mistral AI, has recently unveiled a new large language model (LLM) known as Mistral Large. This new model is set to compete with several top AI systems currently on the market, marking a significant advance over earlier Mistral models. The original article, published on Artificial Intelligence News, provides a comprehensive overview of the new model and its potential impact on the AI landscape.
Mistral AI asserts that Mistral Large has outperformed most major LLMs, with the exception of OpenAI’s recently launched GPT-4, in tests of language understanding. It has also shown strong performance in mathematics and coding assessments. Mistral Large is a proprietary model with fluency in English, French, Spanish, German, and Italian, and boasts a vocabulary exceeding 20,000 words. While Mistral’s first model was open-source, the code for Mistral Large remains closed, similar to systems from OpenAI and other companies.
To enable user interaction with the system, Mistral AI has also launched a chatbot interface named Le Chat, akin to ChatGPT. The launch of Mistral Large and Le Chat comes on the heels of nearly $500 million in funding received by the company late last year from backers such as Nvidia and Andreessen Horowitz. Additionally, Mistral AI has recently partnered with Microsoft to provide access to Mistral Large through Azure cloud services.
The investment of €15 million by Microsoft into Mistral AI is slated to face scrutiny from European Union regulators who are already examining the tech giant’s ties to OpenAI, the maker of market-leading models like GPT-3 and GPT-4. The European Commission has indicated that it will review Microsoft’s deal with Mistral, which could potentially lead to a formal probe jeopardizing the partnership.
Microsoft’s AI efforts have been predominantly focused on OpenAI, with the tech giant having invested around $13 billion into the California-based company. These links are now also under review in both the EU and the UK for potential anti-competitive concerns.
As for pricing, the Mistral Large model starts at $8 per million input tokens and $24 per million output tokens. The system will leverage Azure’s computing infrastructure for training and deployment needs as Mistral AI and Microsoft collaborate on AI research.
While third-party rankings have yet to fully assess Mistral Large, an earlier model by the firm, Mistral Medium, ranked 6th out of over 60 language models. With the release of Mistral Large, Mistral AI appears positioned to challenge dominant players in the increasingly crowded AI space.
