Model-by-model: who overuses what
GPT-5 loves “underscore”, Claude reaches for “nuance”, and everybody overuses “robust”. The per-model fingerprint.
ciopinno··1 min read
Each model has a fingerprint — the words it reaches for more than its peers.
Fingerprints
- GPT-5: underscore, delve, multifaceted.
- Claude: nuance, tapestry, "it's worth noting".
- Everyone: robust, leverage, seamless.
Why it matters
If you mix outputs from two models, the combined fingerprint is a dead giveaway. Edit for one voice.