ICLR 2026 Oral paper introducing FlashWorld, which generates high-quality 3D scenes within seconds from a single image or text prompt. Unlike conventional multi-view-oriented approaches, FlashWorld directly produces 3D Gaussian representations during multi-view generation using a dual-mode pre-training and cross-mode post-training pipeline, achieving 4-7 seconds generation on A100/H100 GPUs.