model roundup

ChatGPT 5.5

3 items · started 2026-05-20 · closed 2026-05-24

  1. TL;DR per chi ha fretta: OpenAI fa A/B testing su utenti senza disclosure (sia free che paid) Uno di questi esperimenti si chiama “things you lose then control” “Things” = utenti. “Lose” = abbandono.

  2. A few weeks ago I randomly opened Chrome’s task manager (I do not recall why I did this exactly) and noticed one tab from my app (ai resume generator) was using over 9.9 GB of memory, and this freaked me out. The worst part was that the is…

  3. Also FYI, Gemini reasoning models (2.5 Pro, 3.0 Pro and 3.1 Pro) were MoE. I don't know why this isn't more broadly discussed.

← all threads