Model inconsistency on weekend temperatures/severe potential - I'm siding with GFS and storms

There are huge discrepancies between the European Centre for Medium-Range Weather Forecast's model (a.k.a. the ECMWF model), and the GFS (operated by the National Weather Service) regarding this weekend's temperatures and severe potential.

SBCAPE for roughly 5PM Saturday according to the ECMWF model is predicted to be less than 500j/kg

The ECMWF essentially cancels any thunderstorm potential for Saturday, and suggests much milder temperatures and dewpoints for Michigan.  This is due to it being much slower in the progression of a storm system moving across southern Canada which looks to bring severe weather to the Dakota's on Thursday.  The models are in agreement regarding the timing and placement of this system, so the SPC has already put out a day 4 Slight Risk for this area.

SPC day 4 categorical outlook highlights central/eastern ND, extreme northeastern SD, and northwestern Minnesota for a 15% risk of severe weather.

However, the GFS tells a much different story for the Great Lakes this weekend.

SBCAPE for roughly 5PM Saturday according to the GFS model is predicted to be over 4000j/kg in some areas

The GFS advances this system much faster, and predicts temperatures in the upper 80s/low 90s with dewpoints in the 70s for Saturday.  This would create massive instability once again, with surface-based CAPE levels rising above 3000j/kg. 

When the models disagree this much, it's difficult to say which scenario will play out, but I'm siding with the GFS's version for a couple of reasons. The first reason is my gut feeling. This past Saturday we saw 90° temperatures and dewpoints in the 70s. It's already happened once, and patterns like this tend to repeat themselves due to the large-scale oscillations at work (NAO, PDO, and everyone's favorite - El Niño, or ENSO). The same oscillations are going to be at work this weekend as were last weekend. The second, more important reason is that the GFS has been far more consistent in depicting this storm system intensify and amplify from run to run.  This leads me to believe it has a more solid grip of the situation than the ECMWF does, which has been somewhat sporadic in regards to the timing and placement, slowing it down even further from its previous run. It helps that the GFS runs four times a day, as opposed to two times a day from the ECMWF, so it's a much easier model to gauge consistency from.

Regardless, as of now I expect a hot, humid weekend with thunderstorms on Saturday, including the possibility of severe storms.  Last night's 8PM run of the GFS was a bit more ominous in this regard than this morning's, with the suggestion that the storm system might amplify considerably and take on a slight negative tilt. For those who don't remember, that means significantly stronger winds, a more powerful cold front, and stronger storms including tornado potential.  More runs will be necessary to gauge this, and for now I agree with the SPC that predictability is just too low to delineate any risk. For now, though, I'd say temperatures should be great for another beach day, though being ready to dodge lightning appears to remain the rule of thumb for Michigan swimmers this summer.