They are both 'right', they just mean two different things. The number returned from Google is not the real link count -- which there is no way to get --but it is a relative measure for comparison between two pages in the rankings. Moreover, since the spider always lags the real web, there will often be pages in the SE's index that no longer link to the target page, but since they are in the index, they do count in the ranking algorithm.

On the other hand, the OptiLink number is the measure right now of the real web, so it could be considered a prospective measure of what the SE index will show once it catches up with changes.

These differences are not terribly important in doing ranking analysis so long as these differences do not predominate. Ranking analysis is by nature both approximate and relative, so errors are acceptable so long as they are not major and distributed more or less equally across the pages being analyzed.