This study aims to produce a model that can meet the needs of Differentiated Learning based on AI. This method used research and development (R&D) with the Generic Desain Research Model (GDRM). The subject of this study was mathematics education experts and mathematics learning model experts. Instruments used in this study include quality validity assessment and feasibility assessment analysis. The data analysis techniques used were quantitative methods with learning model validity assessment criteria that use A Linkert Scale of 1-4. The test results of the AI-based Mathematics learning model show that (1) the Content Validity test resulted in an average of 3.47 which means the model is very valid, and (2) The average feasibility test is 3.43, which means it is also very valid. The results showed that the AI-based Mathematics learning model can be implemented in differentiated mathematics classes and train students to learn independently, think critically, provide solutions to problems, and have the courage to convey solutions.