Overview

• The latest update adds subtle conversational cues to GPT-5, such as brief affirmations, aiming for a warmer, more engaging tone without excessive flattery.
• Paid ChatGPT subscribers regain access to GPT-4 and GPT-4o options following backlash over their removal from the model picker.
• OpenAI introduced four response modes—Auto, Fast, Thinking, and Thinking mini—to let users control the model’s depth and delivery speed.
• Many experienced users characterize GPT-5 as a modest technical upgrade, praising stability but noting limited gains in coding and reasoning tasks.
• Independent benchmark assessments continue to show that some competitor models outperform GPT-5 on accuracy-focused tests.