This is interesting. You would definitely have to 'nail down' the magnets to the sphere, because it will be an unstable configuration. Also in the real world, edge-effects will destroy any chance of perfect radial field lines, so let's assume we're in an ideal scenario.
Outside the sphere, the magnetic field would be that of a source monopole placed at the sphere's center. But we need $\nabla\cdot B=0$, so as a result there is no B-field on the outside.
Inside the sphere, there is nowhere the magnetic field lines can end, especially when they are all pointed towards the center... In fact, such a magnetic field would have divergence less than zero (the center of the sphere being a 'sink'), and this is a property that magnetic fields cannot have (since $\nabla\cdot B=0$). As a result, my answer is that there is no $B$-field on the inside either.
The real reason the B-field must have zero divergence: If there are no physical source monopoles in the vicinity, then any configuration is made of dipoles, and there is no way mathematically (I think) for a collection of dipoles to produce a monopole.