OpenAI has reversed its decision to phase out GPT-4o following widespread user backlash after the launch of GPT-5. Initially, GPT-5 was introduced as a major upgrade to ChatGPT, replacing GPT-4o for most users, with GPT-4o access limited to paid subscribers. The move sparked criticism, particularly on social media and forums, where users expressed dissatisfaction with GPT-5's performance and the removal of GPT-4o. In response, OpenAI restored GPT-4o as the default model for all paying ChatGPT users and reintroduced a model selector feature allowing users to toggle between legacy models such as GPT-4o, o3, and GPT-4.1, as well as GPT-5 variants. OpenAI CEO Sam Altman acknowledged missteps during the GPT-5 rollout and introduced new GPT-5 modes—"Auto," "Fast," and "Thinking"—to provide users with more control over speed and depth of responses. The "Thinking" mode supports a context window of 196,000 tokens and allows up to 3,000 messages per week before switching to a mini version. Additionally, OpenAI is working on personality updates to make GPT-5 warmer and friendlier. These updates and restored access are currently available to paid ChatGPT Plus users, while free users remain limited to GPT-5 without the additional modes or legacy model access. Altman emphasized that OpenAI will provide advance notice if GPT-4o is deprecated again, reflecting a more cautious approach to future changes.