Model cards are proposed as a framework for documenting trained machine learning models to ensure transparency and accountability. The goal is to provide detailed information about model performance across various factors, including cultural, demographic, and phenotypic groups, as well as intersectional analyses. Model cards include sections on model details, intended use, factors, metrics, evaluation data, training data, quantitative analyses, ethical considerations, and caveats and recommendations. They aim to help users understand the strengths and limitations of a model, the types of errors it may make, and how to use the technology more fairly and inclusively. Model cards are intended to accompany released models and provide users with the necessary information to evaluate the suitability of the model for their specific purposes. They also encourage the responsible democratization of machine learning technology by increasing transparency into how well artificial intelligence technology works. The paper provides examples of model cards for a smiling detection model and a toxicity detection model, highlighting the importance of disaggregated evaluation and ethical considerations in model reporting. The framework is applicable to any trained machine learning model and can be used to document the performance of models in various application domains. The paper also discusses the importance of considering intersectional analyses and the challenges of evaluating models across different demographic groups. Model cards are intended to be a standard part of model releases, helping to ensure that models are used in a responsible and ethical manner.Model cards are proposed as a framework for documenting trained machine learning models to ensure transparency and accountability. The goal is to provide detailed information about model performance across various factors, including cultural, demographic, and phenotypic groups, as well as intersectional analyses. Model cards include sections on model details, intended use, factors, metrics, evaluation data, training data, quantitative analyses, ethical considerations, and caveats and recommendations. They aim to help users understand the strengths and limitations of a model, the types of errors it may make, and how to use the technology more fairly and inclusively. Model cards are intended to accompany released models and provide users with the necessary information to evaluate the suitability of the model for their specific purposes. They also encourage the responsible democratization of machine learning technology by increasing transparency into how well artificial intelligence technology works. The paper provides examples of model cards for a smiling detection model and a toxicity detection model, highlighting the importance of disaggregated evaluation and ethical considerations in model reporting. The framework is applicable to any trained machine learning model and can be used to document the performance of models in various application domains. The paper also discusses the importance of considering intersectional analyses and the challenges of evaluating models across different demographic groups. Model cards are intended to be a standard part of model releases, helping to ensure that models are used in a responsible and ethical manner.