Hi and welcome.
A prism is not what you need, imo. (The optics has to be symmetrical about the axis, and a prism only works in one plane. Your idea would actually require a cone and not a prism and it will not produce a focussed spot) What counts is the 'acceptance angle' of the fibre, which is the range of angles of incident light on the end that will actually propagate down the fibre. Whatever the width of the fibre, you need a lens with a focal length that will provide a cone of light that will enter the fibre. The only light that will get into the fibre will be what is within that acceptance cone. If you have a large lens close up, a lot of the light from the spot you have focussed will not get into the fibre. That means you need a long focal length.
You can measure the acceptance angle by observing a small light source and move it from side to side until you can't see it through the fibre or by looking at the cone of light that emerges from the fibre (on a paper sheet) when you put a light at the other end. That will tell you how big your lens needs to be. The longer the focal length, the more efficient the illumination of the fibre.
Note, you can overheat the fibre if the lens area is too great and the power that doesn't get into the fibre will dissipate on the way in. So you should mask off the outer part of the lens to restrict the power input to a safe level.