This paper proposes a new cooperative coevolution (CC) framework for solving large-scale nonseparable optimization problems. The framework introduces a grouping-based decomposition strategy and adaptive weighting to enhance coadaptation among subcomponents. A novel differential evolution algorithm, SaNSDE, is used as the base optimizer for subcomponents. Theoretical analysis shows that the new framework can effectively capture variable interdependencies in nonseparable problems. Extensive computational studies on benchmark functions with up to 1000 dimensions demonstrate that the proposed algorithm, DECC-G, is both effective and efficient for large-scale optimization. The results show that DECC-G outperforms existing algorithms, including FEPCC and DECC-O, on most benchmark functions. It is particularly effective on nonseparable functions and performs well even when the problem dimension increases. The adaptive weighting strategy is shown to be effective in improving the performance of DECC-G, especially for nonseparable functions. The paper also discusses the scalability of DECC-G and its potential for future research.This paper proposes a new cooperative coevolution (CC) framework for solving large-scale nonseparable optimization problems. The framework introduces a grouping-based decomposition strategy and adaptive weighting to enhance coadaptation among subcomponents. A novel differential evolution algorithm, SaNSDE, is used as the base optimizer for subcomponents. Theoretical analysis shows that the new framework can effectively capture variable interdependencies in nonseparable problems. Extensive computational studies on benchmark functions with up to 1000 dimensions demonstrate that the proposed algorithm, DECC-G, is both effective and efficient for large-scale optimization. The results show that DECC-G outperforms existing algorithms, including FEPCC and DECC-O, on most benchmark functions. It is particularly effective on nonseparable functions and performs well even when the problem dimension increases. The adaptive weighting strategy is shown to be effective in improving the performance of DECC-G, especially for nonseparable functions. The paper also discusses the scalability of DECC-G and its potential for future research.