We present Fractional Diffusion Bridge Models (FDBM), a novel generative diffusion bridge framework that enables generative diffusion bridge modeling with fractional noise for both paired and unpaired training data.