Mistral AI is a company that has gained popularity for its expertise in developing foundational AI models. The company was established in May 2023 and has since raised a staggering amount of money in a very short period of time. Within a few weeks of its incorporation, Mistral AI raised $113 million in a seed round. In December, the company successfully secured $415 million in a funding round, with a16z leading the way.
Mistral AI is an AI-focused company founded by alums from Google’s DeepMind and Meta. Initially, the company was known for its open-source approach towards AI development. However, while Mistral AI’s first model was released under an open-source license with access to model weights, the same cannot be said for its larger models.
Presently, Mistral AI’s business model is more aligned with OpenAI’s business model. The company offers Mistral Large through a paid API with usage-based pricing. The charges for utilizing Mistral Large are $8 per million of input tokens and $24 per million of output tokens. This approach seems to be working in favor of Mistral AI, and they are growing their business with sustainable revenue from API usage.
The Paris-based startup has recently announced the launch of their new flagship language model, Mistral Large. This model is designed to compete with other top-tier models like GPT-4 and Claude 2 in terms of reasoning capabilities, and is set to provide an alternative to OpenAI and Anthropic . Along with Mistral Large, the company is also launching a beta version of their new chat assistant called Le Chat, which aims to compete with ChatGPT. Anyone can sign up and try it out on chat.mistral.ai. The company says that it is a beta release for now and that there could be “quirks.”
Access to the service is free (for now) and users can choose between three different models — Mistral Small, Mistral Large and a prototype model that has been designed to be brief and concise called Mistral Next.
In this blog: AI And Human Collaboration In Intellectual Property
Mistral AI is capable of processing large context windows of up to 32k tokens, which can typically cover more than 20,000 words in English. Additionally, the Mistral Large model is equipped to support multiple languages including English, French, Spanish, German and Italian.
Mistral Large’s Multi-lingual Capacities
Possessing inherent multi-lingual capabilities, it significantly surpasses LLaMA 2 70B on various benchmarks such as HellaSwag, Arc Challenge, and MMLU, particularly excelling in evaluations conducted in French, German, Spanish, and Italian languages.
Mistral Large’s Maths & Coding Capabilities
This model has proven to be highly skilled in coding and mathematical tasks, consistently delivering outstanding performance. To provide a thorough evaluation of its coding and math capabilities against other top LLM models, the table below presents its exceptional performance across widely recognized benchmarks.
Partnership with Microsoft
Amidst the announcement for Mistral Large, the Paris based start-up also announced a partnership with Microsoft, which already collaborates with OpenAI. Large is thus available for Azure AI customers, Microsoft’s platform.
Also see: OpenAI Redefining The Future Of Artificial Intelligence
Ending Note
Looking towards the future, Mistral Large is poised to become a leading AI solution for developers and coders seeking advanced machine learning capabilities. With its state-of-the-art architecture, superior accuracy, and flexible customization options, it is becoming increasingly popular among those who seek a powerful and cost-effective AI solution.
As the field of artificial intelligence continues to evolve at a rapid pace, models like Mistral Large will be instrumental in shaping the future of AI technology, unlocking new possibilities and driving innovation across a range of industries. It’s exciting to think about the many ways in which Mistral Large and similar models will transform the world as we know it, paving the way for a smarter, more connected future.