OpenAI recently launched its highly anticipated AI model, GPT-5, but the rollout faced significant backlash from users. Reports surfaced during a Reddit AMA where users criticized GPT-5 for being “dumber” than expected, leading CEO Sam Altman to announce the reintroduction of the older GPT-4o model for ChatGPT Plus subscribers. GPT-5 is designed as a unified version of all prior models, featuring a real-time router intended to optimize responses. However, the router’s initial failure contributed to user dissatisfaction. Additionally, Altman acknowledged a “mega chart screwup” regarding misleading performance benchmarks presented during the launch, which amplified online criticism. OpenAI is now reviewing user feedback to improve GPT-5’s features and transparency about model performance. While legacy models have been removed, Altman assured users that efforts are underway to stabilize GPT-5 and incorporate the desired capabilities, expressing a commitment to ongoing improvement based on user insights.
Source link
Share
Read more