The CAESAR project has developed two QSAR models for assessing developmental toxicity, a complex and challenging endpoint due to the complexity, length, and costs of experiments. The first model, based on a random forest algorithm, and the second, based on an adaptive fuzzy partition algorithm, both performed well. The random forest model has been implemented in the CAESAR online application, a java-based software that allows users to input chemical structures and obtain predictions. The CAESAR models aim to minimize false negatives to make them more useful for REACH legislation. The CAESAR platform ensures easy access and use of the developmental toxicity model by industry and regulators. The models were statistically evaluated using strict criteria and outperformed other models using the same dataset. Tools were developed to guide users toward safer application, such as checking the difference between experimental and predicted values for similar compounds. The CAESAR models and platform address the requirements of REACH, including the need for valid QSAR methods and the assessment of the suitability of each model for specific chemicals.The CAESAR project has developed two QSAR models for assessing developmental toxicity, a complex and challenging endpoint due to the complexity, length, and costs of experiments. The first model, based on a random forest algorithm, and the second, based on an adaptive fuzzy partition algorithm, both performed well. The random forest model has been implemented in the CAESAR online application, a java-based software that allows users to input chemical structures and obtain predictions. The CAESAR models aim to minimize false negatives to make them more useful for REACH legislation. The CAESAR platform ensures easy access and use of the developmental toxicity model by industry and regulators. The models were statistically evaluated using strict criteria and outperformed other models using the same dataset. Tools were developed to guide users toward safer application, such as checking the difference between experimental and predicted values for similar compounds. The CAESAR models and platform address the requirements of REACH, including the need for valid QSAR methods and the assessment of the suitability of each model for specific chemicals.