|Table courtesy of BOM/AU.|
Model forecasts can be found here: http://www.bom.gov.au/climate/ahead/ENSO-summary.shtml
The first category will consist of the first 3 models- the Australian POAMA, the American CFSv2, and the French ARPEGE model. The crucial thing that must be noted with these models is that they started in early September, giving them an advantage over the other models that were started earlier than this. Also, note that these 3 models have a totally different viewpoint on the next 1-3 months, with all models indicating a neutral 1-3 months in terms of ENSO. More monumental is the 4-6 month forecast that covers the winter months. All 3 models now suggest a NEUTRAL winter is in store. Considering underwater cooling is appearing, at the atmosphere remains very stubborn on accepting this El Nino, I can definitely see this as a possibility.
The second category will hold the American GEOS-5 model, the European System 4 (off the ECMWF), the Japanese JMA/MRI-CGCM, and the United Kingdom's GloSea model. First off, note that all of these models started their forecasts on August 1st, meaning that they are at a disadvantage to the first 3 models. These models may not recognize the possibly-occurring downfall of the El Nino, thus flawing their forecasts. The first 1-3 months of these model's forecasts hold a consistent El Nino, but in the 4-6 month forecast, note that all but 1 of these 4 models are showing an El Nino still sticking around. In comparison to the first 3 models, I find these 4 at a significant disadvantage based mainly off when they started their forecast. While you may not see this as a big deal, a month can make a huge difference, and I feel that that is the situation ongoing here.
These are model forecasts, which I have never been too fond of in comparison to analogs and observed conditions. However, based on these observed conditions, I do not find the first 3 models' forecasts to be totally out of line. They are definitely possibilities come winter.