I checked. The information was a little old but the Ethereum testnet got 20-30 per second and they aim to someday have 100k
Really blockchains with low tx/s limits won’t scale. So we want it high.
Imho all tx/s information for most blockchain systems are bullshit and here’ why.
If you perform a a tx right now chances are it’s gonna take somewhere between 1 and 100 seconds to be confirmed (guesstimate).
If we assume a constant 60 sec blocktime then NEM, right now has 120 tx/min. Anything else is bs. You can’t just calculate it to seconds and say 2tx/sec because that’s not how it actually works. Odds are you won’t get 2 tx confirmed in 1 sec.
The per second value should be calculated by checking how many tx will be confirmed in a second if I initiate them right now. The answer to that is most likely 0 unless you are really lucky with the blocktime.
In all traditional blockchain schemes you can only give tx/blocktime values.
@patmast3r: how many tx/sec a blockchain can swallow without dying and how fast txes are confirmed are two different things.
True, but I think when people talk about tx/s they talk about cleared tx not just tx that are in the queue. Could be wrong of course.
When I talk about TPS, it’s the averaged number of transactions the network can handle. In other words the transactions that can be pushed though the system in a sustainable way without creating queues in the long term.
So : average(#transactions per block /block time)
I suppose pat means it’s not an absolute number unless you would have 1 second block times 
Mijin was mentioned in this recent Fast Company article:
MIJIN was recently presented at Tech Crunch Japan in Tokyo.