OpenAI is making significant adjustments to its recently launched GPT-5 model, responding to initial feedback from the user community. The changes focus on increasing usage limits, improving the model's "personality," and bringing back older, popular versions of the chatbot.
Following the release of GPT-5 on August 7, 2025, which replaced previous models like GPT-4o, many users expressed their dissatisfaction. Criticism centered on a more "canned" and less creative demeanor in the new model, as well as tighter usage limits that some felt didn't justify the subscription cost. Some users complained of slower performance and "bland" responses.
In response, OpenAI CEO Sam Altman announced a series of improvements. For GPT Plus subscribers, the weekly message cap for GPT-5's "Thinking" mode has been increased to 3,000. Additionally, users can now choose between three response modes: "Auto," "Fast," and "Thinking," offering greater flexibility. Crucially, the decision was made to restore the GPT-4o model to the model picker for all paid subscribers following the public outcry.
One of the most interesting aspects of the initial GPT-5 announcement was the promise of an improved "personality." The model was designed to be more reliable and professional, acting like a "smart, thoughtful colleague." However, initial user feedback suggested its new nature was perceived as cold. Altman acknowledged this shortcoming and stated the company is working to make GPT-5's personality "warmer," but not as "annoyingly affirmative" as some found GPT-4o to be.
Beyond the immediate fixes, OpenAI has already introduced personality customization options. Users can choose between four preset personalities: "Cynic," "Robot," "Listener," and "Nerd," giving them greater control over the tone of the responses. This feature, known as Custom Instructions, allows users to set preferences that ChatGPT will consider in every conversation.
GPT-5, officially released on August 7, 2025, is a unified system that combines advanced reasoning and multimodal capabilities, designed for complex workflows. Altman described it as a model that can answer easy questions quickly but also "think" for a long time on harder problems. Despite the initial issues, the model is touted for its strong capabilities in coding, math, and information synthesis, with fewer factual errors.
OpenAI's strategy appears to be one of continuous improvement through community interaction. Altman emphasized the company's commitment to transparency and adaptability, acknowledging that the long-term goal is to make ChatGPT "better than it has ever been" for all users. This approach is critical as competition in the AI space from companies like Google, Anthropic, and Meta intensifies.