This informal testing found that Janus Pro explained a Nokia meme much more crisply than DALL-E 3 but was quite a bit worse than the other tasks, even appearing to hallucinate a score in one test case.
I suddenly realize I myself sound like CHatGPT. Haha. Haha.
https://www.analyticsvidhya.com/blog/2025/01/janus-pro-7b-vs-dall-e-3/
This informal testing found that Janus Pro explained a Nokia meme much more crisply than DALL-E 3 but was quite a bit worse than the other tasks, even appearing to hallucinate a score in one test case.
I suddenly realize I myself sound like CHatGPT. Haha. Haha.
Edit: At least you can run these models locally!