This affected 538's national model by 1.3 points. A poll from Oklahoma.
Which is ludicrous, because it moved the Oklahoma odds by much less than that. This is a flaw in the model.
Well in Oklahoma, Trump was already at over 99% so it's not like this poll could move it that far. The idea is that if Trump is further ahead than expected in Oklahoma, he's also more likely to win Ohio and North Carolina and Pennsylvania.
I agree that it's probably a flaw in the model though, due to the overall lack of polling in Oklahoma. This practically cancelled out the Washington Post national poll that came out today, which is silly.