OpenAI is refining its newly launched GPT-5 AI model, aiming for more relatable and warm responses based on user feedback that critiques its previous formality. Upcoming behavioral adjustments, expected next week, will introduce casual affirmations like “Good question,” enhancing user interaction without increasing sycophancy. OpenAI’s CEO Sam Altman acknowledged some rollout mishaps, noting a surge in API traffic but also logistical challenges. Despite this, ChatGPT’s user base has quadrupled, approaching 700 million weekly users. Key updates to GPT-5 include new modes—Auto, Fast, and Thinking—allowing users to choose response speed and depth. OpenAI also reintroduced the GPT-4o model for Plus users and is enhancing usage limits to 3,000 weekly messages for ChatGPT Plus and Team subscribers. Further interface improvements are underway to streamline access to Thinking Mode and clarify model usage. These updates demonstrate OpenAI’s commitment to user satisfaction and effective AI interactions.
Source link
Share
Read more