It would be far easier to write code if you would use cell array for the existing data instead of a bunch of named variables -- but, the way to build the array would be to compute the total length by adding numel() of all to determine the ending array size and allocate it...then place each vector where it belongs in the array.
N1=numel(A);
N2=numel(B);
...
N=N1+N2+...
O=zeros(N,6);
O(1:N1,1)=A;
O(N1+1:N1+N2+1,2)=B;
...
You can see the pattern; if you had all in a cell array of a given name, then you could write a loop that would address each in turn instead of writing out each explicitly...
A={[1 2 3 6 ];
[4 5 5 6 9 8 7 6 9 6 8 6 9];
[4 1 2 3 5 68 9 6 9 6 9 6 9 6 9 63 45 82 85 96 74 52 63 10 30 23 65 6 66 33 66 ];
[5 4 6 5 4 8 9 6 9 6 8 6];
[4 8 9 7 8 6 9 6 8 2 21 3 6 56 36];
[7 8 9 9 6 8 6 45 25 33 66 33 14 25 36 36 12]};
N=cellfun(@numel,A);
B=zeros(sum(N),numel(A));
i1=1;
for i=1:numel(A)
B(i1:i1+N(i)-1,i)=A{i};
i1=i1+N(i);
end
Caution, air code -- untested...
