OLMo is an open-source language model and training framework released by the AI2 Institute. It provides a complete set of resources, including training data, code, model parameters, and evaluation code, enabling researchers to train and experiment with large-scale language models. OLMo's strength lies in its true openness. Users can access complete information from data to models, complemented by comprehensive documentation, facilitating open research and collaboration among researchers. This framework lowers the barrier to entry for language model research, allowing more people to participate and drive advancements in language model technology.