Claude is much funnier than Codex
7:03 PM · Jun 6, 2026 · 7.7K Views
Sentiment
Some users dismissed OpenAI Codex as much worse than Claude at generating funny content and not worth including in the comparison.
Pos
0.0%
Neg
100.0%
2 comments with sentiment.
Cluster Engagement
Posts from X
Most Activity
Most Activity
VIEWS2.9KBOOKMARKS2LIKES7REPLIES1
Miles Brundage@Miles_Brundage
More precisely, Claude Code with high reasoning settings + Opus 4.8 produces funnier content along these lines compared to Codex with GPT-5.5 and high reasoning settings. None of the Codex stuff was worth including in the survey, v. bad
Miles Brundage@Miles_Brundage
Which of these is funniest? (survey in next tweet)
*Warning*: flashing lights in some videos
8hViews 2.9KLikes 7Bookmarks 2
Miles Brundage@Miles_Brundage
Or, "Codex is even less funny than Claude" depending on your perspective
Miles Brundage@Miles_Brundage
More precisely, Claude Code with high reasoning settings + Opus 4.8 produces funnier content along these lines compared to Codex with GPT-5.5 and high reasoning settings. None of the Codex stuff was worth including in the survey, v. bad
8hViews 1.3KLikes 5Bookmarks 1
Jeffcafe, private detective@jeffcafe_
@Miles_Brundage None of them are even remotely funny, I think it’s a good informal benchmark for when AGI getting closer. Your world knowledge and hypothetical reasoning within it to make something funny and original has to be high. Models can only be slightly less unfunny relative to each other
8hViews 7
Miles Brundage@Miles_Brundage
@jeffcafe_ Wherever you want to set the cutoff of funny vs. not funny, the point is, Codex is much worse
8hViews 6