ChatGPT felt dumb on April 13, 2026.

I tried chatting with my early‑stage 100M model and it was a cringe‑fest. Every prompt got garbled replies – “How are you?” turned into “I can I will help with?” and asking the sky’s colour produced a nonsensical sentence. Even simple math like 2 + 2 left the bot clueless. The whole experience felt frustrating and useless, like talking to a broken echo.

I tried using the chat just for fun—asking about hair, makeup, and then curious about how old I look. Every time it refused my request, even when I showed it my own picture. When my husband used the same photo and asked about a woman, it gave a clear answer. The inconsistency felt frustrating and exposed a clear bias, leaving me disappointed in the tool’s behavior.

I switched from the $200 to the $100 plan and immediately felt the Pro model’s power slip away. Tasks that used to keep the model busy for half an hour now stall after ten minutes. My whole workflow—planning, reviewing, design—relied on that deep thinking, and now it’s cut in half. The sudden throttling feels like a crippling downgrade that’s ruining my productivity.

I tried ChatGPT on a bunch of topics. It actually nailed the medical queries when I fed it all the details, which was impressive. But the moment I asked it to solve equations or play word games it started spitting out nonsense words, and a sports question about VJ Edgecombe was flat‑out wrong. Those basic, searchable facts felt embarrassingly off, leaving me frustrated and wondering if there’s a better model for the price I’m paying.

I’ve been using a GPT‑based subreddit bot for a while because its memory and story‑telling were great for fanfiction. Lately the new restrictions have killed my ability to write fight scenes or any killing, which is really frustrating. I’m now looking for a replacement—maybe Grok, Claude, or something else that still offers strong memory and creative storytelling.

I’ve been trying to create RPG characters that only have animal ears, but the AI keeps slipping in human ears despite clear “no human ears” settings. I’ve repeatedly reminded it, even educated it on the rule, yet it just ignores me and adds those human ears the moment I look away. The whole process feels pointless and irritating, and I’m stuck searching for a way to finally stop this persistent mistake.

I set up a 400‑word logic puzzle and ran it through Claude, Gemini, ChatGPT, and Grok, each at both base and pro levels, three times. Every model stumbled—none solved it fully and most botched the simple reading‑comprehension part. The experience was irritating; the AI’s inability to follow the detailed instructions felt like a glaring shortfall in even the newest frontier models.

I asked the chatbot whether “Enjoy Your Trap” by TryHardNinja was copyrighted, but it completely misheard the title and replied with “Enjoy Your Tap.” The response was clearly off‑topic and showed it didn’t understand my request. I felt annoyed that a simple query about a song’s copyright turned into a confusing, incorrect answer, making the interaction feel useless.