InclusionAI's LLaDA-Image-Turbo-FP8 provides the Turbo variant of a model family that generates and edits images. Its model card recommends four sampling steps, compared with 50 for the base variant. Turbo is distillationTraining a small, cheap model to imitate a big one's outputs, keeping much of the capability at a fraction of the cost.Full definition → for this shorter generation process and supports both text prompts and instruction-guided image editing. The step counts describe the generation recipe; they do not establish an equivalent multiplier in measured speed.
Terms in this piece · Glossary
distillation — Training a small, cheap model to imitate a big one's outputs, keeping much of the capability at a fraction of the cost.
inference — Running a trained model to get answers — the phase where AI is actually used, as opposed to trained.
