[MASSIVE RELEASE] Supra2-IMG - a tiny 100M text-to-image model - SOTA quality and open release!

SupraLabs has released Supra2-IMG, a 100M parameter DiT text-to-image model. It was trained from scratch in under 10 hours on a single H100, generating state-of-the-art quality 256x256 pixel images. The model can be run locally, taking approximately 20 seconds per image on CPU and 2 seconds on GPU. Users can download an inference script and generate images with specified prompts, seeds, CFG, and steps.