Mistral AI startup introduces LLM, challenging ChatGPT, Gemini, and Claude.

France-based startup Mistral AI has introduced its new proprietary large language model (LLM), Mistral Large, which aims to compete with major players in the AI market. The company claims that Mistral Large performed well in a “multitask language understanding” test, outperforming several notable LLMs, with the exception of GPT-4. Additionally, Mistral Large excelled in math and coding tests. However, the model has not been compared to xAI’s Grok and Google’s Gemini Ultra, which were released recently. Mistral AI has stated that Mistral Large is significantly better than its earlier models. The company has also unveiled “Le Chat,” an AI chat interface built on top of its models, similar to ChatGPT’s structure. Mistral Large is fluent in multiple languages and has knowledge of over 20,000 English words. The model is closed and proprietary, in line with other recent LLM releases by OpenAI. Mistral AI recently secured $487 million in funding from companies such as Nvidia, Salesforce, and Andreessen Horowitz. Mistral Large is yet to be evaluated by third-party platforms, but Mistral Medium, the previous model, ranks sixth out of more than 60 LLMs on Chatbot Arena. Mistral AI has also formed a partnership with Microsoft, making Mistral Large accessible on Azure AI Studio and Azure Machine Learning. The model will utilize Azure’s supercomputing infrastructure for training and scaling purposes, and the two companies will collaborate on AI research and development. Mistral Large is priced at $8 per one million tokens of input and $24 per million tokens of output, making it slightly cheaper than GPT-4 Turbo. Mistral AI was valued at approximately $2 billion in December.