They're merely different drag models used for predicting ballistic performance of a given bullet. The G1 is the "standard" bullet used throughout the industry, despite the fact that it's a relatively poor fit for most of our more modern projectiles. It's relatively short (about three calibers in length), flat based, and has a rather blunt ogive. In the absence of anything denoting some other drag model, you can usually correctly assume that a manufacturer's listed BC is based on the G1 model. Berger (or more specifically, Bryan Litz) has lead the charge to use the G7 drag model for the more modern designs. The G7 model is significant;y more streamlined, has a boat tail and more accurately represents the flight characteristics of modern VLD or OTM Match bullets we use today. The result of this is that a ballistic program using the G1 model for a trajectory calculation will see some significant variation downrange, from what is calculated. The farther downrange, the more that discrepancy will grow. For a bullet more similarly shaped to the G7 model, the calculated trajectories will be far more accurate when compared to the actual firing results.
There's several other drag models in this same series (E.D. Lowry's, developed in the late 1940's or early 1950's) that may be more appropriate for some other given bullet, say, a flat-nose .30-30 design or a LSWC pistol bullet, but they see very little use outside some professional circles. Hope that helps a bit, but believe me, this is a very (VERY) abbreviated explanation of a rather complex topic.