User Protests Work! OpenAI Reverts to Old Models, CEO Hints at GPT-5 Pro Trial Access

OpenAI’s aggressive rollout of GPT-5 and removal of older models sparked intense backlash from users worldwide. Amid the protests, OpenAI quickly adjusted its strategy, not only restoring the GPT-4o option but also having CEO Sam Altman hint at offering Plus subscribers a trial of GPT-5 Pro. This is a victory for the users.

Sometimes, a tech giant’s decision is like a high-speed train—clear destination, but it might just ignore the scenery and the passengers’ feelings along the way. Recently, OpenAI experienced firsthand how an “emergency brake” could win back applause.

When some users found their ChatGPT interface forcibly upgraded to GPT-5, with familiar models like GPT-4o gone, an unexpected storm began to brew. Social media and developer forums were filled with complaints, with many users lamenting that their essential workflows were completely disrupted.

“GPT-5 is powerful, but for certain tasks, it’s not as stable as GPT-4o!” “Why take away our choice?"—similar voices rose, creating immense public pressure.

Clearly, OpenAI underestimated users’ reliance on and emotional connection to the older models. Faced with a torrent of negative feedback, the AI giant made a wise choice: listen, then compromise.

A Victory for the Users: How to Get Your GPT-4o Back

After realizing that a “one-size-fits-all” upgrade strategy wouldn’t work, OpenAI acted swiftly to restore access to older models. This wasn’t just a technical adjustment; it was an apology and a sign of respect for the community.

If you want to find that familiar GPT-4o, the process is very simple:

1. Go to Settings: Click on your account name in the bottom-left corner of the ChatGPT interface and go to “Settings.”
2. Enable the Legacy Switch: In the “General” tab, find the “Display legacy models” option and make sure its switch is on.
3. All Set: Return to the main screen, and in the model selection dropdown, you’ll see a “Legacy models” category. Click it, and GPT-4o will be waiting for you.

This seemingly small change is actually significant. It shows that OpenAI acknowledges that on the road to more powerful AI, user habits and the right to choose are equally important.

Not Just a Compromise, but an Olive Branch of Goodwill

Appeasing user dissatisfaction was just the first step. Next, OpenAI seems prepared to show even greater sincerity.

Remember what OpenAI CEO Sam Altman initially said? He mentioned considering offering Plus subscribers a “(very) small number” of GPT-5 Pro queries as a sneak peek. But after this incident, his stance on the matter seems to have become more positive.

In a later exchange, he told a developer willing to pay $1,000 a month for higher input limits: “Feels like we should find a way to enable this…” This statement was widely interpreted as a strong signal: OpenAI is seriously considering offering a more attractive GPT-5 Pro usage plan for its paying users.

This is actually a very clever strategy. As a “research-grade” top-tier model, GPT-5 Pro’s power is undeniable. Instead of letting the Pro version be out of reach, it’s better to generously let all paying users experience its charm. Once users personally feel the efficiency gains from GPT-5 Pro, the true heavy users and enterprise customers will naturally be more willing to pay for it. This is both an “olive branch” to soothe users and a brilliant market education.

Re-understanding the GPT-5 Family: Why Is Choice So Important?

This incident also made more people realize that different AI models have their own suitable scenarios, and it’s not just a simple matter of “new versus old.”

• GPT-5 (Flagship Model): As the new default model, it is undoubtedly the most powerful in general capabilities, suitable for handling the vast majority of daily tasks.
• GPT-4o (Previous Mainstay): The controversy proved that it still has irreplaceable advantages in stability, speed, and cost-effectiveness for certain specific tasks.
• GPT-5 Thinking: When you need deep, detailed, and insightful answers, this “thinker” model is the best choice.
• GPT-5 Pro: For professional-grade, research-level tasks that require extremely high precision and complex reasoning, the Pro version is where its true strength lies.

Forcing everyone to use the same model is like making everyone wear the same size shoe—it’s obviously not appropriate. OpenAI’s compromise this time is a return to the service essence of “providing different tools for different needs.”

In the end, this policy shift, sparked by user protests, is perhaps a better outcome for everyone. It reminds all tech companies that even when at the forefront of technology, they must never ignore the voices of their users. And for users, it proves that active, rational feedback can indeed push products to become more user-friendly and flexible.

Now, the choice is back in our hands. Will you continue to embrace the all-powerful GPT-5, or return to the familiar embrace of GPT-4o? Or, like me, are you looking forward to the arrival of that legendary GPT-5 Pro trial opportunity?