This report is still reasonably strong, except for the record inventory and months of inventory.
Update: Greg Ip, writing in the WSJ's Washington Wire, points out that due to a change in the sample, housing price data will be biased downward:
Tricky Housing Trends: New-home sale prices are always tricky to analyze because they are heavily influenced by the mix of homes sold: more luxury home sales will tend to bias up the figure, more homes sold in the south or Midwest, where prices are lower, will bias it down. But in the last year an additional issue has muddied the trend. In January 2005, for the first time since 1985, the Census Bureau updated the sample of local permit offices it checks to track new-home construction and prices. A lot changed between 1985 and 2005. Older, pricier areas became heavily built up and activity declined. In newer, outlying areas, where prices were generally lower, construction picked up. The 2005 sample thus has a larger share of those newer, cheaper areas. Comparing 2005 figures to those in 2004 will give the impression of a slowing in price gains, but that's somewhat artificial.