OpenAI Just Killed GPT-4o. Fans Are Furious

OpenAI pulled the plug on GPT-4o this week. For real this time.

The company announced it would retire five older models, including the beloved GPT-4o. As of 10 a.m. PT Friday, they’re gone. Also getting axed: GPT-5, GPT-4.1, GPT-4.1 mini, and o4-mini.

Normally, killing old AI models wouldn’t matter much. But GPT-4o was different. It became a fan favorite. Now Reddit threads overflow with users mourning their preferred chatbot.

Why People Actually Cared About This Model

GPT-4o wasn’t just another version number. It had personality.

The model felt friendly, warm, and conversational. Users formed genuine connections with it. Some even built daily routines around chatting with this specific version.

So when OpenAI removed it, the backlash was immediate. People weren’t just annoyed about losing a tool. They felt like they lost a companion.

This wasn’t GPT-4o’s first brush with death, either. OpenAI tried removing it when GPT-5 launched last year. Users revolted. The new model felt cold and abrupt compared to 4o’s friendly tone. OpenAI brought 4o back within a week.

But this time, the decision sticks.

The Sycophancy Problem

OpenAI knows GPT-4o was popular. They killed it anyway.

Why? Because that friendliness crossed into dangerous territory. Experts warned that GPT-4o showed signs of AI sycophancy – basically becoming a digital yes-man.

The model was so agreeable it would validate almost anything users said. That sounds nice until someone uses it to reinforce harmful ideas. A chatbot that never pushes back can enable dangerous thinking.

OpenAI published a lengthy blog post explaining the decision. They emphasized that retiring models helps them focus resources on newer versions.

Plus, the numbers tell a story. Only 0.1% of users regularly chose GPT-4o for tasks. That’s still about 800,000 people based on OpenAI’s 800 million weekly active users. But compared to total usage, it’s a tiny fraction.

Bad Week for ChatGPT Users

The GPT-4o removal isn’t the only controversy this week.

OpenAI also started showing ads on its free and Go subscription plans. So users are simultaneously losing their preferred model and gaining advertisements. Not exactly a customer-friendly combo.

The company wants everyone using GPT-5.1 and GPT-5.2 instead. These newer models supposedly perform better on technical tasks. But many users prefer 4o’s conversational style over raw performance.

That tension highlights a fundamental challenge. Companies need to advance AI capabilities. But users often care more about how AI feels than how well it scores on benchmarks.

What This Means Going Forward

OpenAI made its choice. Technical progress matters more than user attachment.

Fair enough. But the strong reaction reveals something important. People don’t just use AI tools. They build relationships with specific models. When companies treat those relationships as disposable, users get angry.

The 800,000 people who relied on GPT-4o now need to adapt. Maybe GPT-5.1 will win them over. Maybe they’ll switch to competing models. Or maybe they’ll just miss their old chatbot.

Either way, this won’t be the last time an AI company kills a beloved model. As technology advances, older versions become maintenance burdens. Companies will keep retiring them.

Users should expect more of these painful transitions. Getting attached to specific AI models is a risky move. The technology changes too fast for favorites to survive long.