After the negative reactions to GPT 5, we may see model versioning that asymptotically approaches the next whole number without ever reaching it. "New for 2030: Claude 4.9.2!"
replyOr approaching a magic number like e (Metafont) or π (TeX).
replythe standard used to be that major version means a new base model / full retrain... but now it is arbitrary i guess
reply