Leaked Fable 5.1 system prompt compared against ChatGPT in behavioral test

doodlestein · x · 2026-09-02

A post compares the output of ChatGPT against a prompt excerpt derived from the recently leaked Fable 5.1 system prompt. The comparison highlights differences in response styles and tendencies, sparking discussion about model alignment and jailbreaking behaviors. Such tests are often used to evaluate variations in safety guardrails and instruction following across different models.

Related event: Leaked Fable 5.1 System Prompt Sparks Claude vs ChatGPT Comparisons(2 posts)→

Original post →

More from Models

Models channel →