OpenAI unveiled its highly anticipated GPT-5 during a live-stream on Thursday, touting major strides in reasoning, writing, coding, accuracy, health-related queries, and a sharp reduction in “hallucinations” — instances where the AI produces incorrect information. The model began rolling out almost immediately after the announcement, replacing several older OpenAI models in the process.

However, less than a day later, platforms such as Reddit and X (formerly Twitter) were awash with criticism. Many social media users expressed frustration over losing access to their preferred models, with discussions on online forums suggesting that GPT-5 delivers briefer responses and lacks the personality of earlier versions. Regular users who previously spent hours engaging with the chatbot also complained that, even with a ChatGPT Plus subscription, they now have fewer prompts available.

Online backlash over GPT-5 launch

A Reddit thread titled ‘GPT-5 is horrible’ attracted nearly 3.1K upvotes and more than 1,300 comments, most of them negative.

One user, headwaterscarto, mocked the launch demo, saying, “I like how the demo they were like - “if it gets something wrong no worries, just ask again. I’m actually going to run 3 prompts at once and pick my favorite” like how is that better???? I’m feel like i’m taking crazy pills”

Another, NutzHang, agreed by saying, “Agreed. Answers are shorter and, so far, not any better than previous models. Combine that with more restrictive usage, and it feels like a downgrade branded as the new hotness.”

User GERRROONNNNIIMMOOOO likened it to “an OpenAI version of shrinkflation,” while AnonRep2345 argued that old models should have remained available until GPT-5 was refined.

Some lamented disrupted workflows. JacckSparow wrote, “For months I was in perfect sync, switching between o3, o3-pro, 4.5, and 4o depending on the task, knowing exactly what each model could deliver. Now they’re suddenly gone, and I’m stuck readjusting to GPT-5, which is already throwing off my flow. Tried it just now and it butchered my job description. I work in marketing, and it says I “handle voice & image.” Seriously? How the heck does the smartest model answer like this??”

User SnooSketches6133 went further by saying, “GPT-5 is not an upgrade. It is a forced downgrade. GPT-5 hallucinates. It forgets context. It feels unstable and inconsistent. GPT-4o was sharp, focused, and reliable. It actually worked. I did not subscribe to be part of a forced A/B test with no way out. I want GPT-4o back. I want the option to choose. @OpenAI, stop breaking what people trusted. Bring back GPT-4o.”

Criticism also poured in on X. One user complained about “three minutes to think about a response” followed by a stream error, pleading for GPT-4’s return.

Another user accused the new model of mixing up informationsaying, “theres seriously something wrong with Chatgpt. its mixing up info, doesnt give the right info anymore, not even simple yes or no question, like a literal stupid robot which it is bruh. But I had better experiences before. idk whats goin on with chat... disappointed.”

Another reported being unable to download files, describing GPT-5 as “fully broken” and “really, really slow.” He said, “GPT5 is fully broken right now. I cant download any file, its really really slow, I ask it to do something trivial and easy and i get a reply that isnt even anything to do with what i asked. I also keep getting hella error in stream messages @OpenAI”