This OpenAI research presents DALL-E 3: revolutionizing text-to-image models with enhanced cue-tracking capabilities 10/31/2023