GPT 5.4 and 5.3 codex behavior inconsistent in ways to far from conventional llm behavior to be normal

Resolved 💬 3 comments Opened Mar 7, 2026 by ludoplex Closed Mar 7, 2026

💡 Likely answer: A maintainer (github-actions[bot], contributor) responded on this thread — see the highlighted reply below.

Review session and llm behavior for yourself and see if you cannot identify behavior far outside of conventional llm generation of these model versions. Will provide full transcripts when they are verified as not containing sensitive info, and once sensitive info redaction completes.

View original on GitHub ↗

3 Comments

github-actions[bot] contributor · 5 months ago

Potential duplicates detected. Please review them and close your issue if it is a duplicate.

#13823
#13114
#13260
#11966

Powered by Codex Action

etraut-openai contributor · 5 months ago

Thanks for the bug report. We'll share the feedback with the team responsible for training our codex models. If you see other model behavior that you'd like to report, please use the /feedback command. That will provide us with additional details including logs and conversation text.

ludoplex · 5 months ago

Thanks for the bug report. We'll share the feedback with the team responsible for training our codex models. If you see other model behavior that you'd like to report, please use the /feedback command. That will provide us with additional details including logs and conversation text.

Dont close the issue as resolved or I will open another one, you'll want to see what the LLM has done and how far outside usual injection sabotage it actually is as it is unprecedented