I am pretty sure that OH map would point to a Trump win.
Almost certainly.
Then why would their model have Clinton ahead? That doesn't make sense--their numbers are based on that map. Just because a map that looks like that might have historically been a R win doesn't mean it would in this election. It's completely reasonable (in fact, exactly as expected) to assume that rural/suburban counties would shift towards Trump while urban counties would shift towards Clinton, meaning fewer "blue" counties but a similar overall margin.
This isn't an estimate of the final vote, it's an estimate of the current vote