It's rare that a model introduced in a paper has fair comparisons. You'll want to see it used by many different authors before one can speak with confidence on claimed improvements.

In this paper, Bellard makes a comparison to the Transformer, some variant of which holds most NLP records. Of note is that in this test, the Transformer did not outperform the LSTM. Is it that Transformers need to be larger to outperform or LSTMs are more suited to compression or that more epochs are required? This deserves further investigation, what are the required conditions for Transformers to underperform LSTMs?