Mistral partners with Microsoft, announces new large AI model with chat app

In MMLU tests, Mistral Large had an accuracy of 81.2%, sitting behind GPT-4’s, but doing much better than Llama 2 70B and Gemini Pro 1.0.

70b accuracy ai ai model amazon amazon bedrock app bedrock business chat conversational ai foundation-models gemini gemini pro generative-ai gpt gpt-4 large language models llama llama 2 microsoft mistral mistral ai mistral large mistral large model ml and deep learning mmlu multilingual ai multilingual models nlp partners tests