OpenAI’s rollout of GPT‑5 has drawn mixed reactions, with early user feedback pointing to noticeable drops in performance compared to its predecessor. CEO Sam Altman acknowledged the issues in a recent Reddit AMA, attributing the degraded experience to malfunctions in the new real-time model router, a core feature designed to dynamically direct prompts to the most suitable engine. The router’s instability left some users feeling as though GPT‑5 was less capable than GPT‑4o, prompting OpenAI to explore restoring GPT‑4o access for certain users.
The company also faced public scrutiny over a misleading data visualization shown during its launch livestream, where a performance comparison chart sparked accusations of misrepresentation. OpenAI later corrected the graphic and admitted the error, describing it as unintentional. In response to growing concerns, the company has temporarily doubled the usage limits for ChatGPT Plus subscribers, providing additional capacity as the team works to stabilize routing behavior and improve model transparency. While OpenAI maintains that GPT‑5 offers significant advancements, the rollout has revealed the challenges of deploying complex AI systems at scale—particularly when user expectations are shaped by prior models with distinct behaviors.
