OpenAI’s new GPT-5 model, launched as a unified replacement for older GPT and o-series models, is drawing backlash from ChatGPT users. Many say it feels less capable, especially compared to GPT-4o, which they relied on for creativity, nuance, and long-form interaction. Complaints cite shorter, less engaging responses and the loss of 4o’s personality.
In a Reddit AMA, CEO Sam Altman acknowledged the concerns, blaming an “autoswitcher” glitch that stopped GPT-5 from matching the right capability to each query. He promised fixes, higher rate limits for Plus users, and temporary continued access to GPT-4o.
OpenAI faces a dilemma — keeping older models adds to already heavy losses, but removing them frustrates loyal users. Last year, the company made $3.7B in revenue but lost $5B, much of it on server costs.