Widespread Criticism Hits GPT-5 as Users Demand Return of Previous Models
OpenAI’s latest artificial intelligence model, GPT-5, made its official debut on Thursday with promises of being a major leap forward in AI capabilities. According to CEO Sam Altman, the model is designed to deliver superior reasoning, improved writing quality, and more accurate programming assistance compared to previous iterations. The launch immediately rolled out to all ChatGPT users, with free accounts receiving limited access.
Despite these ambitious claims, the reaction from the AI community has been far from celebratory. Within just 24 hours of release, social media platforms—particularly Reddit—were flooded with critical feedback. Many long-time ChatGPT users described GPT-5 as “underwhelming” and even “a downgrade” from its predecessors, especially when compared to GPT-4.0 and GPT-4.1. Sentiments like “I miss 4.1, bring it back” and “They should have let us keep the old models until the new one was fixed” have been widely shared.
One of the key points of frustration among ChatGPT Plus subscribers is the removal of several popular AI models from the platform. Before GPT-5’s arrival, paid users had access to a variety of reasoning models such as o4-mini, o4-mini-high, and o3. These options have now been replaced with a single alternative—GPT-5 Thinking—which comes with a weekly usage cap of just 200 messages. For many paying customers, this feels like a reduction in value rather than an upgrade.
OpenAI maintains that GPT-5 offers advanced capabilities, particularly in reasoning-intensive tasks, claiming it can “think” more effectively when necessary. However, the company’s messaging, amplified by Altman’s public statements, has been accused of overselling the benefits and underdelivering on performance. Several AI experts have noted that while GPT-5 does show improvements in certain programming scenarios, its conversational abilities feel less natural and sometimes less accurate than GPT-4.1.
The controversy also highlights a broader challenge in AI product development: balancing innovation with user expectations. As models become more complex, even small changes in tone, speed, or accuracy can have a significant impact on user satisfaction. For many, the issue is not only about GPT-5’s capabilities but also about losing access to familiar tools they had relied on for months.
Looking ahead, some speculate that OpenAI may introduce a hybrid access system, allowing users to choose between older and newer models based on their preferences and needs. Until then, the debate over GPT-5’s merits—and whether it truly represents a leap forward—remains a hot topic in the AI community.