I don't know those numbers offhand. SR5 work is loaded manually and with 98 k values remaining, we eliminate k's often enough that I don't generate new work until it's needed. In fact the last time I loaded SR5 work there were 100 k values left and then in the week after that work was loaded two SR5 primes were found.. In researching this, I see that I never eliminated those two k values from the sieve. I'm going to do that now.

Afterwards, the only way I really have of knowing how many candidates are in any given range is to open the sieve (candidate) file in an editor and count lines. But the thing is that we're almost certainly going to eliminate more k values before all that work is completed.

(later edit)

We're currently loaded to n=2.9M

n=2.9M-3.0M has 69081 candidates
n=3.0M-3.1M has 69341 candidates
n=3.1M-3.2M has 69275 candidates

You can get a good enough approximation from that. The count doesn't vary much.

(even later edit)

Each candidate becomes a workunit, each workunit needs two matching tasks to validate. So double the numbers above if you want a task (result) count.