Generate images from text prompts using multiple models