Stable Zero123 is an in-house trained model for view-conditioned image generation. Compared to the previous state-of-the-art model Zero123-XL, Stable Zero123 produces significantly improved results. It achieves this through three key innovations: 1. A substantially filtered training dataset from Objaverse, retaining only high-quality 3D objects and rendering them more realistically than previous methods. 2. An estimated camera angle is provided to the model during both training and inference, enabling it to make more informed and higher-quality predictions. 3. Pre-computed datasets (pre-calculated latent variables) and an improved data loader that supports higher batch sizes, coupled with the first innovation, result in a 40x improvement in training efficiency compared to Zero123-XL. The model is now available on Hugging Face for researchers and non-commercial users to download and experiment with.