Efficiency of lights and quads vs sprites

I am working on a project where our artist wants to have a gradient light on the background. We can either tile strips of the background and apply a light source using quads, or we can have a single giant background with the gradient baked in, but it gets to be a sprite. The whole rest of the scene is full of sprites already - so my question is (since we’re developing on mobile), which is more efficient?

Quads and meshes inherently sound more expensive to me, since they must be 3D, while sprites sound like they would be cheaper. At the same time, I know one huge background sprite is a bad idea.

In addition: let’s say we just had tiled sprites, how much of an improvement would that be over the tiled quads with lighting?

We have these two options within the artist’s style, and then one other option that breaks that for the sake of efficiency. I’m wondering how I can find out how much each factor matters.

Sprites are 3D; they are quads. (Unless you have Pro, in which case they can optionally conform to the shape of the sprite, which uses more polygons but saves on fill rate.) I don’t know what a “gradient light” is, but anything involving lights is inherently more expensive than not using lights. Anything with transparency is inherently more (potentially much more) expensive than not using transparency.

–Eric