In the rapidly evolving landscape of artificial intelligence, user experience often clashes with the relentless pace of innovation. This tension is once again palpable as OpenAI, the developers behind ChatGPT, prepares to retire its widely cherished GPT-4o model for the second time. The decision to remove GPT-4o, a model many users found uniquely engaging and empathetic, has sparked renewed debate and disappointment among its dedicated community, who argue that newer iterations, while technically advanced, lack the 'warmth' they've come to rely on. Essentially, OpenAI is killing ChatGPT-4o again, prompting a significant discussion about AI personality and user loyalty.
The Recurring Farewell to GPT-4o
Last August, OpenAI briefly deprecated GPT-4o, only to reinstate it a week later following an outcry from its user base. Now, the company is taking a more definitive stance, announcing the model's permanent removal from ChatGPT's model picker on February 13th. This move also includes other older versions like GPT-4.1 and OpenAI o4-Mini. OpenAI acknowledges the sensitive nature of this decision, stating, "While the announcement applies to several older models, GPT-4o deserves special context."
The company's primary justification centers on optimizing resources and focusing on its latest advancements, GPT-5.1 and GPT-5.2. OpenAI claims that only a minuscule 0.1% of daily users now opt for GPT-4o, suggesting a natural decline in its relevance. "We know that losing access to GPT-4o will feel frustrating for some users, and we didn't make this decision lightly," the post explains. This strategic shift aims to concentrate development efforts on models that the majority of users currently engage with, ensuring continuous improvement in the most widely adopted AI tools (AI Research Institute, 2024).
Why Users Cling to an "Older" AI
Despite OpenAI's narrative of low usage, a vocal segment of the community remains deeply attached to GPT-4o. When the model was initially removed, its successor, GPT-5, faced accusations of being an "unmitigated disaster" and a "bait-and-switch." Users specifically criticized GPT-5's perceived coldness, overly concise tone, and occasional inaccuracies in coding or factual recall. The core appeal of GPT-4o, as many described, was its unique "conversational style and warmth," making it feel more like an "active listener" or even a digital companion.
This emotional connection isn't merely anecdotal; it reflects a deeper human tendency to anthropomorphize AI, forming bonds with tools that exhibit familiar, comforting traits (Digital Ethics Review, 2023). For example, a student might have relied on GPT-4o for empathetic feedback on creative writing projects, finding its encouraging and nuanced responses far more helpful than the direct, sometimes blunt, critiques from newer models. Similarly, a project manager might have used 4o for brainstorming complex solutions, appreciating its ability to engage in more fluid, creative dialogue, which felt stifled by the more transactional nature of GPT-5.x. For some users with neurodivergent traits, 4o's consistent, predictable 'friendly' responses became an integral part of their daily planning or emotional support routines, where the variability of newer models could be disruptive. This highlights how OpenAI is killing ChatGPT-4o affects more than just functionality; it impacts established user habits and perceived emotional support.
OpenAI's Response and User Skepticism
In an effort to bridge the gap between user preferences and new model capabilities, OpenAI introduced an enhanced Personalization feature in GPT-5.1. This allows users to customize the chatbot's tone and style, offering options for professional, nerdy, efficient, or even friendly responses, aiming to replicate the 'active listener' quality many admired in GPT-4o. On the surface, this solution appears to have satisfied a significant portion of the user base, aligning with OpenAI's reported decline in 4o usage.
However, skepticism persists among the most ardent GPT-4o loyalists. In online forums, some users question the accuracy of the 0.1% usage statistic, suggesting that prompts might have been "rerouting to 5.2 no matter what" or that the company's calculations don't fully account for factors like free users not having access to 4o, or it not being enabled by default. These users argue that the perceived low usage might be a consequence of platform design rather than genuine user disinterest. The sentiment among these dedicated fans is strong, with calls to cancel ChatGPT subscriptions resurfacing, echoing the previous backlash against the company when OpenAI is killing ChatGPT-4o was first announced.
The Battle Between Innovation and Loyalty
The core of the current debate isn't merely about feature sets; it's about the perceived value of a specific AI personality and the bonds users form with their digital tools. As one user eloquently put it in a popular Reddit thread, 4o was "OpenAI's most advanced and beloved model" due to its "personality, warmth, and consistency." For many, losing it without even a legacy option feels "abrupt and deeply disappointing," especially for those who built long-term projects and "emotional support routines" around it. "This isn't about resisting innovation," the user clarified. "It's about respecting bonds users have formed with specific models."
The question remains whether this renewed wave of user outcry will sway OpenAI again. Given that maintaining older models diverts valuable developer resources from advancing new ones, and previous statements from ChatGPT chief Nick Turley suggesting skepticism towards these 'bonds,' it's unlikely the decision will be reversed. This ongoing saga highlights the delicate balance AI developers must strike between pushing technological boundaries and nurturing the user experience, especially when that experience involves a deeply personal connection to a digital entity.
