By realmnova 
OpenAI officially released the GPT-5 model series earlier and promptly distributed it to all users. While OpenAI stressed that GPT-5 had made a major breakthrough in reasoning, mandate implementation and multi-modular understanding (e.g., image, audio and text processing), some users were not satisfied with the GPT-5 experience and faced the announcement by Sam Altman, chief executive officer of the disputed company, that he would be back on the GPT-4o line.
The release of GPT-5 is defined by OpenAI as the “Next Generation of Generating AI Bar”. Official data show that GPT-5’s accuracy in mathematical reasoning (e.g., AIME testing) has increased to 92 per cent (GPT-4o, 83 per cent), code generation efficiency has increased by 30 per cent and support for multimodular tasks (e.g., image description and audio processing) has increased by 50 per cent. Its training data are three times larger than GPT-4o and, combined with more efficient reasoning algorithms, significantly enhances the ability to handle complex tasks.
However, there are senior users who provide feedback on social platforms in their use, GPS-5 is too “mechanized” for dialogue and lacks the emotional temperature of GPT-4o, such as in creative writing, where stories generated by GPT-5 lack narrative depth. Some users also found that work processes based on GPT-4o optimization (e.g. Prompt Project) needed to be significantly adjusted in the new model, affecting production efficiency.
Free users are dissatisfied with the increased use restrictions. The high-calculative demand of GPT-5 has led to a reduction in the number of free user dialogues from 50 to 30 per day from GPT-4o, with users spitting on X and “free users scrubbing the amount in 10 minutes, forcing the upgrade of ChatGPT Plus ($20 per month)”.
OpenAI’s GPT-5 post has made a breakthrough in reasoning and multi-modular capabilities, but user dissatisfaction with interactive experiences and free amounts highlights the challenge of balancing technology with user needs. In response, Altman responded at Reddit AMA that the computing cost of GPT-5 was twice that of GPT-4o, but that OpenAI would optimize free user experience and restore the GPS-4o option for ChatGPT Plus users, with the supporting time frame adjusted to the user feedback dynamics.
