This is an AI image generation comparison for
text-to-image 
prompt:
Will Smith meme reads "I hate spaghetti", as he is shown screaming and throwing a bowl with spaghetti back at the viewer refusing to eat it. Bowl and spaghetti along with tomato sauce and meatballs are captured mid flight in the mid-background and foregraund. Dramatic, epic, comically exagerated...
Log in to see full prompt.
Tested: April 26, 2025
ChatGPT's new image model can make memes straight up. Here's mine.
Tested: April 11, 2025
Celebrity image and text generated by the Dev model from Huggingface space
Tested: April 24, 2025
Dreamina Seedream 3.0 doing a meme with text. In all 4 generated images it kept the taxt in the exact case it was supplied, without inferring that a meme usually means all caps, so keep that in mind if you want all caps - enter that in your quotes.
Tested: April 26, 2025
"I HATE SPAGHETTI" supplied in all caps
Tested: August 7, 2025
Generated in cartoonish style since there was no style reference, but overall pretty great. I'll do photorealistic next
Tested: August 7, 2025
This is a great one. Add 'photorealistic, cinematic' keywords to ensure the realistic style
Tested: October 17, 2025
Tested: October 17, 2025
Makes very lively images for this scene, text isn't correct though
Tested: November 26, 2025
It's uncensored regarding celebrities. For this prompt, it refused to make the subject face the viewer even when specifically asked, he's looking at the bowl
Tested: January 14, 2026
That's definitely not Will, GLM-Image! AND he looks like he loves spaghetti while I've prompted him refusing to eat that and throwing it at the viewer
Tested: April 16, 2026
No Will Smith but typical meme font understanding is great
Check out the results from ChatGPT's Image Generator by OpenAI (GPT‑4o) vs HiDream-I1 vs Dreamina AI (Seedream 3.0) vs Reve vs Qwen Image & Video Generator (Qwen-Image) vs Qwen Image & Video Generator (Qwen-Image) vs Reve (Reve Image 1.0) vs Fal AI (HunyuanImage 3.0) vs Hugging Face (Z-Image Turbo) vs WaveSpeedAI (GLM-Image) vs Hugging Face (ERNIE-Image) for similar or identical prompts side-by-side.
