AFAIK it's pretty common for new macs to spend a while creating an index of its hard drive. For that reason, if you want to run benchmarks, you should generally wait until it's done with that (e.g. an hour or probably less with these speedybois). It might be that the people running the Pro benchmarks didn't wait for that, in their rush to publish the first benchmark. This would be consistent with what we're seeing - the Pro has faster single core performance, but slightly lower multicore, because some of its "background" cores were busy creating the index, while the Air was done with that task.