The paper introduces the Amortized Generative 3D Gaussians (AGG) framework, which aims to generate 3D objects from a single image using an amortized pipeline. Unlike existing methods that require per-instance optimization, AGG instantly produces 3D Gaussians, eliminating the need for computationally expensive score-distillation steps. The framework decomposes the generation of 3D Gaussian locations and appearance attributes into two distinct networks, allowing for joint optimization. It also includes a cascaded pipeline that first generates a coarse representation and then up samples it with a 3D Gaussian super-resolution module. The method is evaluated against existing optimization-based 3D Gaussian frameworks and sampling-based pipelines, demonstrating competitive performance in both qualitative and quantitative metrics while being significantly faster. The project page is available at <https://irld.github.io/AGG/>.The paper introduces the Amortized Generative 3D Gaussians (AGG) framework, which aims to generate 3D objects from a single image using an amortized pipeline. Unlike existing methods that require per-instance optimization, AGG instantly produces 3D Gaussians, eliminating the need for computationally expensive score-distillation steps. The framework decomposes the generation of 3D Gaussian locations and appearance attributes into two distinct networks, allowing for joint optimization. It also includes a cascaded pipeline that first generates a coarse representation and then up samples it with a 3D Gaussian super-resolution module. The method is evaluated against existing optimization-based 3D Gaussian frameworks and sampling-based pipelines, demonstrating competitive performance in both qualitative and quantitative metrics while being significantly faster. The project page is available at <https://irld.github.io/AGG/>.