The paper introduces AlphaFLOW and ESMFLOW, a novel approach that combines AlphaFold and ESMFold with flow-matching to generate protein ensembles. The method leverages the accuracy of single-state predictors and fine-tunes them under a custom flow matching framework to create sequence-conditioned generative models of protein structures. These models, AlphaFLOW and ESMFLOW, are trained on the Protein Data Bank (PDB) and further fine-tuned on molecular dynamics (MD) simulations to capture conformational flexibility and higher-order ensemble observables. The evaluation shows that AlphaFLOW and ESMFLOW outperform existing methods in terms of precision and diversity, providing superior results compared to AlphaFold with MSA subsampling. Additionally, AlphaFLOW can diversify static PDB structures with faster convergence to equilibrium properties than replicate MD simulations, making it a valuable tool for studying protein dynamics and conformational changes. The code for this method is available at <https://github.com/bjing2016/alphaflow>.The paper introduces AlphaFLOW and ESMFLOW, a novel approach that combines AlphaFold and ESMFold with flow-matching to generate protein ensembles. The method leverages the accuracy of single-state predictors and fine-tunes them under a custom flow matching framework to create sequence-conditioned generative models of protein structures. These models, AlphaFLOW and ESMFLOW, are trained on the Protein Data Bank (PDB) and further fine-tuned on molecular dynamics (MD) simulations to capture conformational flexibility and higher-order ensemble observables. The evaluation shows that AlphaFLOW and ESMFLOW outperform existing methods in terms of precision and diversity, providing superior results compared to AlphaFold with MSA subsampling. Additionally, AlphaFLOW can diversify static PDB structures with faster convergence to equilibrium properties than replicate MD simulations, making it a valuable tool for studying protein dynamics and conformational changes. The code for this method is available at <https://github.com/bjing2016/alphaflow>.