I say effective hashrate as opposed to "actual" hashrate. Rejects are low. It's just a matter of how hashrates are calculated by the pool. Multipool explains it well: "The hashrate your miner displays is its own internal hashrate for calculating difficulty 1 shares, and thus is fairly meaningless to the pool. The pool displays an average hashrate calculated based on your valid, accepted shares per minute."
5250/5550 = ~95% which isn't bad considering it's connected through a stratum proxy for leasing.
I would say that is acceptable. If the network connection is slow on the pool, the numbers get lower. And you get some Job not found rejects.
I found a problem in the code now when the hashcounter sum get's higher than 2^32. for a 16000MHASH rig, the rig will hash fine for 5 mins, and then might run into trouble. Testing now at suchpool.pw.