This paper introduces diffusion bridge implicit models (DBIMs) for accelerating the sampling process of denoising diffusion bridge models (DDBMs) without additional training. DDBMs are a powerful variant of diffusion models for interpolating between two arbitrary paired distributions, but their sampling process is computationally intensive. DBIMs generalize DDBMs by defining a class of non-Markovian diffusion bridges on discretized time steps, maintaining the same marginal distributions and training objectives. This generalization allows for a range of sampling procedures, from stochastic to deterministic, and results in faster sampling, up to 25 times faster than vanilla DDBMs. DBIMs also induce a novel form of ordinary differential equations (ODEs) that inspire high-order numerical solvers, further enhancing efficiency. The initial step of DBIMs introduces a booting noise to maintain generation diversity, enabling faithful encoding, reconstruction, and semantic interpolation in image translation tasks. Experiments on high-resolution datasets and image inpainting tasks demonstrate the superior sample quality and efficiency of DBIMs.This paper introduces diffusion bridge implicit models (DBIMs) for accelerating the sampling process of denoising diffusion bridge models (DDBMs) without additional training. DDBMs are a powerful variant of diffusion models for interpolating between two arbitrary paired distributions, but their sampling process is computationally intensive. DBIMs generalize DDBMs by defining a class of non-Markovian diffusion bridges on discretized time steps, maintaining the same marginal distributions and training objectives. This generalization allows for a range of sampling procedures, from stochastic to deterministic, and results in faster sampling, up to 25 times faster than vanilla DDBMs. DBIMs also induce a novel form of ordinary differential equations (ODEs) that inspire high-order numerical solvers, further enhancing efficiency. The initial step of DBIMs introduces a booting noise to maintain generation diversity, enabling faithful encoding, reconstruction, and semantic interpolation in image translation tasks. Experiments on high-resolution datasets and image inpainting tasks demonstrate the superior sample quality and efficiency of DBIMs.