Comment by lifty

1 day ago

I made this switch months ago, ChatGPT 5.4 being a smarter model, but I’ve had subjective feelings of degradation even on 5.4 lately. There’s a lot of growth in usage right now so not sure what kind of optimizations their doing at both companies

Agreed. Watching the intermediate "Thinking about X ... Now I'll do Y" text on GPT 5.4 lately has been like watching a hypothetical smart drug wear off.

All of the major models have been getting worse lately, not just Opus.

  • Makes me wonder if the output is starting to get back into the training input and we're seeing the first signs of model collapse.