That would depend on the probability of a beneficial mutation occurring. Which is a difficult thing to estimate, as most experimental tests of the question only address ‘beneficial in a specific artificial environment’, rather than any possible beneficial mutation. And ignores the fact that mutations without any adaptive effect in one environment may be beneficial in another.
Still, I doubt the rate of beneficial mutations is any lower than 1 every 10k generations or so, since the experimental values are higher in nearly every experiment I’m familiar with. So if ‘reasonable probability’ means 5%, then 500-600 generations. If it means 95%, then ~30k generations. ~280k generations since the most recent common ancestor with chimpanzees, so I’d say the probability of at least one beneficial mutation is really good.
If you disagree, and…
…let me know.