Everyone seems to keep looking at IHME. I am just throwing this out there for a tad bit of variety, as there are numerous models and the IHME is only one.
I am partial to this one, the one generated by the team here at Los Alamos. Admittedly some bias is probably at play because I know some of these people. Anyway, here it is:
https://covid-19.bsvgateway.org/
It reports the 90% confidence interval, the "average" numbers for the predictions, and as a new prediction is generated weekly, it also tells you how the current prediction for the next week are tracking (I.E., best case, worst case, about average, or "wow, we were way off").
It doesn't presume to know how people will behave when restrictions begin to get lifted, or even what restrictions will be lifted and when. So it runs a sort of Monte Carlo simulation whereby the different possible situations are randomized and then takes the average and reports on the 90% range of all outcomes. This is why the Los Alamos model has one of the largest ranges if you look at longer term predictions. But I still think it is a more realistic and appropriate model because if you are reporting a tight confidence band three months out, it means you are assuming an awful lot of facts not in evidence about what will happen and how people will behave and also how well the virus spreads during summer. And I think it is probably a better one for short and medium term forecasts of up to 6 weeks or so. Also, you can look through the plots and see that the 70% confidence interval is much tighter than the 90% interval.
In short, it has huge confidence intervals for the long term because honestly nobody has a clue what the landscape will look like in August and anyone who says they do is selling snake oil.
Throwing this out there since there is so much "Models are crap" attitude. They (mostly) aren't. But they are poorly understood, and I think this one may be easier to 'get' than most.