The CAESAR project developed two QSAR models for developmental toxicity using different statistical methods. The first model uses a random forest algorithm, while the second uses an adaptive fuzzy partition algorithm. Both models performed well, with the random forest model achieving 84% accuracy in testing and the adaptive fuzzy partition model achieving 88% accuracy. These models were implemented in the CAESAR online application, which is a Java-based tool allowing users to access and use the models freely. The CAESAR models aim to minimize false negatives to improve usability for REACH. The application ensures that both industry and regulators can easily access and use the developmental toxicity model, as well as models for other endpoints. The models were validated using an external test set, with the random forest model showing better performance than previous studies. The CAESAR platform provides a web-based interface for users to submit chemical structures, calculate descriptors, and obtain predictions. The platform also includes tools for evaluating the applicability domain of the models, ensuring the reliability of predictions. The models use bi-dimensional descriptors, as tri-dimensional descriptors did not yield better results. The CAESAR project also organized a workshop to discuss the results and improve the models based on user feedback. The models were developed with a focus on scientific validity and reproducibility, and the platform ensures that users can access and use the models effectively. The models were validated using a dataset of 292 compounds, with the results showing good performance compared to other studies. The CAESAR platform provides a user-friendly interface for accessing and using the models, and the models are available for free. The models were developed to meet the requirements of the REACH legislation, which requires the assessment of chemicals for various endpoints, including developmental toxicity. The CAESAR models aim to provide reliable and accurate predictions for developmental toxicity, helping to reduce the need for animal testing. The models were developed using a combination of statistical methods and chemical descriptors, and the platform ensures that users can access and use the models effectively. The models were validated using an external test set, with the results showing good performance compared to other studies. The CAESAR platform provides a user-friendly interface for accessing and using the models, and the models are available for free. The models were developed to meet the requirements of the REACH legislation, which requires the assessment of chemicals for various endpoints, including developmental toxicity. The CAESAR models aim to provide reliable and accurate predictions for developmental toxicity, helping to reduce the need for animal testing.The CAESAR project developed two QSAR models for developmental toxicity using different statistical methods. The first model uses a random forest algorithm, while the second uses an adaptive fuzzy partition algorithm. Both models performed well, with the random forest model achieving 84% accuracy in testing and the adaptive fuzzy partition model achieving 88% accuracy. These models were implemented in the CAESAR online application, which is a Java-based tool allowing users to access and use the models freely. The CAESAR models aim to minimize false negatives to improve usability for REACH. The application ensures that both industry and regulators can easily access and use the developmental toxicity model, as well as models for other endpoints. The models were validated using an external test set, with the random forest model showing better performance than previous studies. The CAESAR platform provides a web-based interface for users to submit chemical structures, calculate descriptors, and obtain predictions. The platform also includes tools for evaluating the applicability domain of the models, ensuring the reliability of predictions. The models use bi-dimensional descriptors, as tri-dimensional descriptors did not yield better results. The CAESAR project also organized a workshop to discuss the results and improve the models based on user feedback. The models were developed with a focus on scientific validity and reproducibility, and the platform ensures that users can access and use the models effectively. The models were validated using a dataset of 292 compounds, with the results showing good performance compared to other studies. The CAESAR platform provides a user-friendly interface for accessing and using the models, and the models are available for free. The models were developed to meet the requirements of the REACH legislation, which requires the assessment of chemicals for various endpoints, including developmental toxicity. The CAESAR models aim to provide reliable and accurate predictions for developmental toxicity, helping to reduce the need for animal testing. The models were developed using a combination of statistical methods and chemical descriptors, and the platform ensures that users can access and use the models effectively. The models were validated using an external test set, with the results showing good performance compared to other studies. The CAESAR platform provides a user-friendly interface for accessing and using the models, and the models are available for free. The models were developed to meet the requirements of the REACH legislation, which requires the assessment of chemicals for various endpoints, including developmental toxicity. The CAESAR models aim to provide reliable and accurate predictions for developmental toxicity, helping to reduce the need for animal testing.