Why use PyMol for modelling? It's more about visualising the interactions of molecules alone.

For modelling, you can use any other desktop-based or web-based software.

You can use MODELLER software, it is freely available. You have to just input the sequence and the output will be the modelled structures. Based on the secondary structure analysis of the protein, you can accept the modelled structure as one of your desired structure.

Pymol is not designed to do the modelling, it is used to visualise the result of the modelling process and to compare different models. If you are not familiar with modelling, you best use a Webserver to generate a 3D model of your sequence.

Before starting on modelling, check whether an experimental structure of your protein or a close relative is available by doing a sequence search on the Protein Database (https://www.rcsb.org/search/advanced)

Swissmodel ( https://swissmodel.expasy.org ) is a good place to start, because the "Search for Templates" function gives you a very good overview of the available experimental structures on which your model will be based, and allows you to see whether these span large conformational differences, e.g. dependent on oligomerization or ligand binding, therefore allowing you to decide whether a single model is sufficient to fully describe the conformation of your protein. Look at the detailed modelling reports, as it also gives you information about subsequences predicted to be highly flexible or natively unfolded, ligands, complexes with other proteins or oligomerization state of the template proteins

For sequences that prove difficult to align to potential template sequences, iTasser ( https://zhanggroup.org/I-TASSER/ ) may be a better choice, as it is not dependent on a correct sequence-to-sequence alignment, but directly determines how well your sequence fits the template structure

The most advanced of the structure prediction programs is AlphaFold (e.g. https://colab.research.google.com/github/deepmind/alphafold/blob/main/notebooks/AlphaFold.ipynb) - however, as a novice modeller, be aware that it also does include in the model predictions for sequences that in reality do not assume a defined structure, and whether other factors such as ligands, oligomerization or a membrane separating the extracellular and intracellular part of a protein will influence the native structure. Alphafold-DB https://alphafold.ebi.ac.uk will give you access to pre-built models for the human proteome

Try out different methods, look in detail at the modelling reports generated (e.g. information on the level of reliability of different parts of the model) and compare the different model to the nearest experimental structures.

You can use pymol for visualization and generating high quality figures. For homology modelling, you can use MODELLER and its use is straight forward or you can use an online server as SWISS-MODEL. You can also find some homology models available for some classes of proteins like in the GPCR database.