On Monday, Mistral AI, the French 10 months old rising star in the field of Generative Artificial Intelligence (GenAI), made waves with a significant announcement. The company unveiled its highly anticipated ChatGPT competitor along with a series of new models, marking a bold step forward in the realm of GenAI. What's particularly noteworthy about this announcement is the backing it received from tech giant Microsoft through a $16 Mio investment according to TechCrunch, showing the growing interest and investment in GenAI in Europe.
Mistral AI's foray in the Generative AI space with its rivals OpenAI and Anthropic comes at a time when demand for advanced Generative AI solutions is at an all-time high. With businesses increasingly turning to AI-driven Co-pilots in many of the products they use - notably Microsoft's own O365 products, Gen AI chatbots to streamline customer service and automating various tasks like we do at Structize AI, the competition in the GenAI algorithm space is clearly on.
Here are my key takeaways from the latest Mistral AI model Feb 2024 update when using them in Structize AI. We tested all 3 of them: Mistral Small, Mistral Medium & Mistral Large.
Mistral Small: Only real open-source and fastest Mistral Small emerges as the standout among Mistral AI's offerings, being the only truly open-source model in their lineup. While it may lack the sheer power of its larger counterparts, Mistral Small compensates with speed. However, I have been encountering occasional issues, such as missing output for lengthy requests.
Mistral Medium: No real update & stability issues remain I was not able to find any major updates while testing it. Despite its potential to be a good 'in-between' of Mistral Small and Large, Mistral Medium still falls short in terms of reliability, 60% of my requests resulted into a timeout. Still considerable work on the plate here.
Mistral Large: Performance slight better than GPT 3.5 but costly Positioned as a competitor to established models like GPT-3.5 & GPT 4.0, Mistral Large offers commendable performance, albeit falling short of the capabilities of GPT-4.0. However, its slower processing speed and higher cost may deter some users, with the model's primary appeal lying in its compliance with EU data regulations.
In summary, while Mistral AI's recent announcement marks a significant milestone in the company's journey, it's evident that there are still areas for improvement and refinement. Despite the promising advancements showcased in their new models, established players like OpenAI, with their GPT series, continue to reign supreme in the realm of Generative AI.
Nevertheless, Mistral AI's collaboration with industry heavyweight Microsoft underscores the company's potential to carve out a niche in this competitive landscape, driving innovation and pushing the boundaries of what's possible in GenAI. We should also not forget that Mistral AI has only entered into the arena about 10 months ago. In retrospect, it is quite impressive of what they have been able to achieve in such a short timeframe. We believe that Mistral AI is going to increase its appeal to EU-based companies that want to ensure EU data residency.
Can't wait to get started with Mistral AI or other GenAI models?
You can get started with Mistral AI or other GenAI models using Structize AI to structize your business processes. Book your intake here for €99.-. Not satisfied? We have a 10-day money back guarantee.