There are a number of programs/webservers for protein structure predictions. Which one works best depends on how closely your sequence is related to the nearest experimentally determined structure. First of all you ensure that the structure is not already known - do a sequence similarity search on the PDB: https://www.rcsb.org/search/advanced . If there are clear homologs, Swissmodel will do a good job ( https://swissmodel.expasy.org ). For more distantly related proteins, where the correct sequence alignment is hard to discern, you might want to use a threading-based algorithm, e.g. iTasser https://zhanggroup.org/I-TASSER/. Your protein may also have already been predicted using an AI-based predictor, Alphafold https://alphafold.ebi.ac.uk, or you can submit your sequence for AlphaFold prediction (https://colab.research.google.com/github/deepmind/alphafold/blob/main/notebooks/AlphaFold.ipynb )
The optimal program to use may also be informed by the purpose of your model: For example, if you plan to perform docking /virtual screening on you model, you would preferentially go for template-based modelling on liganded template structures to ensure that the ligand binding site is open and accessible in your model, and you want to compare multiple model to get an idea of the structural flexibility of the protein.