OpenAI tweaks GPT-5 to make it ‘warmer and less annoying’
After a wave of complaints about GPT-5, OpenAI CEO Sam Altman has promised improvements. The AI should become warmer, less distant, and more fun to use again, compared to the previous version, GPT-4o. Users will also get more control over the chatbot’s personality.
Since its launch, Reddit and other platforms have been flooded with complaints. Many users felt that GPT-5 gave shorter and less engaging answers, even to simple questions. The model was also said to be slower and allowed fewer prompts – even for paying ChatGPT Plus users. Some fans suspect OpenAI deliberately shortened answers to reduce costs.
Advanced users especially lamented the loss of the ‘conversational style’ that made GPT-4o so popular. According to them, GPT-5 is more structured, but lacks charm and spontaneity. The upgrades, they say, barely provide any real improvement.
To address these concerns, OpenAI is introducing three new modes: Auto, Fast, and Reflection. This gives users the choice between speed and depth. The default setting remains Auto. The extensive Reflection mode, however, comes with a limit of 3,000 messages per week – after reaching it, you switch to a ‘mini-reflection’ mode.
Responding to community feedback, GPT-4o is once again available to all paying users. OpenAI promises to provide timely warnings should the model be withdrawn again. Subscribers can also enable alternative models like o3, 4.1, and GPT-5 Thinking mini via the ‘Show extra templates’ option. GPT-4.5 remains exclusive to Pro customers.
Business AM
Subscribers 0
Fans 0
Followers 0
Followers