Mistral AI Launches New LLM and Chatbot to Compete with OpenAI

Mistral AI, a France-based firm, has introduced a new proprietary large language model (LLM) to an increasingly congested AI industry, claiming that its new Mistral Large can outperform several key competitors.

In a post on February 26, the Paris-based startup stated that Mistral Large surpassed many major LLMs in a "multitask language understanding" test, except GPT-4, and fared well in numerous maths and coding tests.

However, Mistral Large's performance was not compared to that of xAI's Grok or Google's Gemini Ultra, both released in November and early February, respectively. Cointelegraph has reached out to Mistral AI for comment.

Guillaume Lample, founder and principal scientist, believes Mistral Large is "vastly superior" to Mistral AI's previous models. Mistral AI also released "Le Chat" – an AI conversation interface built on top of its models, similar to how ChatGPT is based on GPT-3.5 and GPT.

The company, which received US$487 million in investment in December from Nvidia, Salesforce, and Andreessen Horowitz, stated that Mistral Large is proficient in French, Spanish, German, and Italian.

While Mistral AI's initial model was provided under an open-source license, Mistral Large is a closed, proprietary model, akin to OpenAI's current LLMs, which has disappointed some watchers on X.

While third-party AI chatbot rating sites such as Chatbot Arena have not evaluated Mistral Large, its previous model Mistral Medium ranks sixth out of over 60 LLMs.

The hundreds of pairwise scores in Chatbot Arena are processed via a Bradley-Terry model, which uses random sampling to provide an "Elo" rating, estimating which model is most likely to win in direct competition with another model.

The company has just announced a relationship with Microsoft, which would make Mistral Large available on Azure AI Studio and Azure Machine Learning.

"Microsoft's trust in our model is a step forward in our journey," Mistral AI stated of its commercial LLM.

Mistral Large will use Azure's "supercomputing infrastructure" for training and scaling, and the two companies will also work on AI research and development, according to a Feb. Eric Boyd, corporate vice president of Microsoft's Azure AI Platform, announced on the 26th.

Mistral Large costs US$8 per one million tokens of input and US$24 per million tokens of output, making it marginally less expensive than GPT-4 Turbo at US$10 and US$30, respectively. Mistral was worth roughly US$2 billion in December, according to Bloomberg.