The Flair English POS tagger is one of the state of the art models that is commonly available.

It’s based upon Flair embeddings and LSTM-CRF.

In my experience, it outperforms SpaCy en_core_web_trf, but this may depend upon your dataset. However, it is slow unless you have a GPU.

Hence, this replicate model.

Please cite the following paper when using this model.

@inproceedings{akbik2018coling,
  title={Contextual String Embeddings for Sequence Labeling},
  author={Akbik, Alan and Blythe, Duncan and Vollgraf, Roland},
  booktitle = {{COLING} 2018, 27th International Conference on Computational Linguistics},
  pages     = {1638--1649},
  year      = {2018}
}