I was going off HuffPo for Sanders beating it by 5 points. RCP makes it even clearer how much Sanders outperformed his polling.
For reference, the most recent Nevada poll we had was from the amateur organization Overtime Politics, which was 47-44 Clinton on Jan 22nd. While amateur, they got 55-41 Sanders-Clinton in New Hampshire and 48-47 Sanders-Clinton in Iowa, which made them actually one of the most accurate pollsters for both.
If you want to disclude that, the last poll we had from Nevada was Gravis from the 27th of December, where it was 50-27 Clinton-Sanders, for a 23 point lead. Two points about that: firstly, that's from December. Sanders is polling about 6 points higher nationally now than he was then, and conversely Clinton 6 points lower. If we assume Nevada follows national trends (and this is very likely because Nevada is quite close to the average Democratic primary demographically), that would make it 44-33 Clinton-Sanders. Secondly, Gravis has hugely overstated Clinton in all their polls. In New Hampshire, they overestimated Clinton by 4 and underestimated Sanders by 2. In Iowa, they overestimated Clinton by 3 and underestimated Sanders by 7. In New Hampshire, they overestimated Clinton by 4 and underestimated Sanders by 14. Even if we assume they make a same-sized mistake as Iowa, that would make it 41-40 Clinton-Sanders. Accounting for Don't Knows reducing now we're in February, those are very similar numbers to Overtime Politics.
If you want to exclude Gravis because they're shit, the last poll we had from Nevada was CNN/ORC from the 10th of October (yes, Nevada is really badly polled). They had it 50-34 Clinton-Sanders. Since then, Clinton has gained 5 and Sanders 8 nationally (Biden was still in the race at that point). That would make it 55-42. In New Hampshire, CNN/ORC overstated Clinton by 5 and understated Sanders by 9. That would make it 50-49 Sanders-Clinton. Again, this is in a very similar ballpark to both of the prior two when you account for differences in Don't Know.
My guesstimation is that Nevada is actually very close right now. That's not just me thinking that. Jon Ralston, the leading Nevadan political analyst thinks so. Sanders has been quietly picking up some key Nevadan and Arizonan Hispanic community leaders. Rumour has it that both campaigns' internal polling is picking up a steady swing in the Hispanic vote - not enough for Sanders to lead the Hispanic vote, but a small enough gap to make the state very competitive.
The big downside is that Sanders has essentially no ground game there, so there are big questions as to whether he can capitalize on this momentum in a caucus state.
If Sanders wins Nevada, all bets are off. He is my favourite for the nomination at that point, no doubt about it. If the margin is in within 5 points, we're looking at a long race that will be genuinely competitive as far as the Convention, especially with Sanders fundraising like it is. If the margin is 5-10 points, Clinton is the overwhelming favourite but the race will probably still continue. If the margin is greater than 10 points, everyone gets off the Bernie bus and starts cheering for the Hilldawg.
That's my call. Nevada is critically important for this race. Let's see how it goes!