I know he spent serious time on it. I don't know how many hours per day. But it is worth noting that he started with a good memory, had learned several languages before then and has studied mnemonic techniques. I doubt I could sustain a rate anything like his.
He was learning Japanese while living in Canada, so I'm pretty sure that he doesn't need an immersive context for that rate. However I doubt he could pick up on the finer nuances without being in an immersive context.
Also, I wonder if he needs to be in an immersive context for that rate to be sustained.