OpenAI’s CEO, Sam Altman, acknowledged that GPT-5’s performance has underwhelmed, particularly in creative writing and reasoning capabilities, despite its strengths in coding and speed. Users reported GPT-5 often produces overly concise responses. Altman attributed these issues to a bug affecting the model’s autoswitcher, which compromised its reasoning functionality. To address user concerns, OpenAI plans to double rate limits for GPT Plus users, allowing them to revert to legacy models like 4o if preferred. A new update will roll out soon to enhance GPT-5’s reasoning abilities and introduce a user-friendly toggle to trigger these capabilities. Although the model is being gradually released to both paid and free users, full access may take time. Simultaneously, OpenAI is focusing on restoring older model options for Pro subscription holders, ensuring continued support amid these updates.
Source link
