![Mistral AI Unveils Mistral Large 2: A Powerful New Language Model](/content/images/size/w600/2024/07/MistralAI-large2.jpg)
Mistral AI Unveils Mistral Large 2: A Powerful New Language Model
On benchmarks such as HumanEval and MultiPL-E, Mistral Large 2 outperforms comparable models, including Llama 3.1 405B instruct, and scores just below GPT-4o. It also demonstrates strong performance on the MATH benchmark, falling only slightly behind GPT-4o.