The technical report on Stable Diffusion 3 (SD3) provides a detailed overview of the multimodal diffusion Transformer architecture, MMDiT, used by SD3. This architecture enhances performance by employing separate sets of weights for image and text representations. The report also reveals the introduction of the reweighting stream technique in SD3 and discusses the scalability studies for future performance improvements. Additionally, the report addresses issues with the text encoder and offers recommendations. Overall, the technical innovations and performance of SD3 leave a profound impression.