Kolors is a cutting-edge text-to-image model powered by latent diffusion. Trained on billions of pairs, it excels in visual quality, complex semantics, and text rendering, outperforming both open and closed-source models.