Google's PaLM 2 is an advanced large language model designed to excel at reasoning tasks such as math, classification, multilingual proficiency, and natural language generation. PaLM 2 outperforms its predecessor through the use of compute-optimal scaling, a more diverse pre-training dataset mixture, and updated model architecture and objectives. The model has been evaluated for potential harms and biases, and demonstrates improved multilingual capabilities, with the ability to code in languages like Python and JavaScript, and generate specialized code in languages like Prolog, Fortran, and Verilog.