Vibeleaderboard
← All Intel
Intel / video

Shaping model behavior in GPT-5.1— the OpenAI Podcast Ep. 11

Source
youtube.com
Author
OpenAI
Date

Context

Two OpenAI staff, a post-training research lead and a product manager for model behavior, describe on the company's podcast what changed in GPT-5.1 after users said GPT-5 had weaker intuition and felt less warm. They say that with 5.1 every model in ChatGPT is a , meaning one that decides how long to think before answering.

They say the coldness came partly from the system around the model. The carried too little of what a user had said earlier, so the model could seem to forget a bad day after ten turns, and an automatic switcher between chat and reasoning models changed the response style mid-conversation. The changes they describe were carrying context further, following custom instructions more consistently, and adding style and trait options for personality. This is OpenAI staff describing their own work; no user results are given.

Terms in this piece · Glossary
  • reasoning model — A model trained to think — generating extended internal reasoning before answering — trading time and tokens for accuracy on hard problems.
  • context window — The maximum amount of text a model can consider at once — its working memory for the current conversation or task.
Read the source www.youtube.com
More from OpenAI
Recommended reads
Comments

Checking sign-in…

Loading comments…