[SleekEagle]

It can run on higher end consumer grade GPUs at a batch size of 1, but there are already forks making it more efficient, so the landscape is evolving very quickly right now!