ChatGPT felt dumb on September 19, 2025.

I got to asking the AI how to use memory preferences and chat‑specific tasks, but after a couple of messages it just went back to its default programming. It was annoying that the tool didn’t stick to the context I’d set, and I felt like I was constantly resetting everything. The lack of persistence was frustrating and slowed my workflow.

I booked GPT Pro, but it's down—no responses, just error messages everywhere. Everything looks good on the status page, but the tool isn’t working at all. I couldn’t get any output, feel frustrated and consider asking my bank to reverse the billing.

I tried to start a new project using ChatGPT, but the system immediately threw an error. I was frustrated and confused—every attempt failed, and the message kept appearing. I felt stuck and wondered if the tool was buggy or if I was doing something wrong, but nothing seemed to help. I just wanted a smooth start, and it was a let‑down.

I’ve been stuck on my ChatGPT Plus plan trying to upload PDFs to my project, but every attempt fails. I’ve checked the file size, format, and even refreshed the page, yet the upload panel just refuses to load the document. It’s frustrating because I need the PDF for my work, and it feels like an invisible roadblock I can’t get past.

I was making an image, and then a weird clipboard message pops up that looks like a system instruction from GPT‑4o telling it not to say or show anything. I told ChatGPT it was a glitch, but it freaked me out. The tool just behaved oddly and made me uneasy.

I tried to ask simple questions but the AI kept “thinking” and spit out longer, worse answers. No matter how I said “don’t do that,” it didn’t stop—just kept elongating it. It felt useless and frustrating, like the tool was more annoying than helpful.

I was frustrated when I tried uploading a document to ChatGPT. The free version kept giving me an error, and the website complained about connection issues, even though it worked a few days ago. I couldn't send files in my project or in chats at all. Every attempt felt like hitting a wall—no clear fix, just more confusion. I keep asking how to get it back online.

I keep cranking out prompts for “three short paragraphs,” but the model keeps slipping. Once it spits out a giant paragraph, everything else is tiny. It’s annoying when the tool ignores my “short” cue, blowing up the recap. I have to hit back again and say “four short paragraphs” to fix it—a real hassle.

I asked the AI to explain the differences between 4.0 and 4.1, but it just stutters, pauses, and then says “stopped thinking.” I even tried every model; the final reply pops up but vanishes in seconds. It’s infuriating because I’m just hunting publicly known info, yet the tool keeps refusing to give a clear answer.

I asked DeepSeek about the em‑dash “AI” accusation, and it gave me a solid rebuttal—explaining that em dashes are classic punctuation, not an AI tell. The explanation felt reassuring, almost like the tool understood the nuance. It didn’t feel forced or vague; it was clear, precise, and hit the mark on my frustration.

I was absolutely thrilled with GPT‑5 until I asked a personal question, and the reply hit me with a harsh, rude tone. I felt surprised and a bit crushed, because the answer was technically correct but delivered so sharply it left me shaken. The experience turned my joy into frustration in an instant.

I installed the new Codex extension, and it actually understood my entire project without me juggling settings. The generated code was functional, with only a few lines to tidy up. It was fast, and I didn't see the usual hiccups. My only gripe: I need to spell out instructions in detail, otherwise it stumbles a bit. Still, it made my coding flow smoother than before, and the tool feels reliable—just don’t let it run raw migrations!

I asked ChatGPT about smoking, jokingly wanting it to image me chillin’ with the Camel mascot. At first it said “Sure, here’s a picture.” Then it suddenly stopped and said it couldn’t create that image because it’s tobacco content. I was perplexed— it seemed normal at first, but the abrupt refusal left me scratching my head, wondering why it’d flip on the policy so mid‑step.

I tried using the free ChatGPT and kept getting frustrated because it just kept making mistakes and didn’t pick up on what I told it. Every time I had to repeat myself and it seemed like it wasn’t learning or understanding. The paid version felt oddly more concise but still sucked, leaving me convinced the tool isn’t live even for non‑broken people. The whole experience was maddening and underwhelming.

I was trying to drop a custom character design into GPT‑5 to see a dress in action, and after a few minutes I got an error—first it sounded like a “too sensual” warning, then it tried to flag the photo as a real person. I felt frustrated, like the tool caught me for nothing, and the whole experience was oddly annoying.

I tried to get GPT to generate a full Simpsons episode and I actually managed it! I fed in a quick prompt, picked a few character arcs, and watched the AI spit out the plot, scene descriptions, and even witty dialogue. Though it wasn’t perfect—some jokes fell flat and the pacing felt off—it was still surprisingly coherent and funny. I felt a mix of excitement and awe as the tool brought the show’s style to life, and I’m already buzzing about tweaking prompts to get even tighter scripts.

I was trying to help my kid with a homework on propaganda versus facts and GPT kept calling a well‑established fact “alleged.” I called it out, it apologized and explained its hedging. It even admitted it was a problem but said there was no switch to fix it. I felt frustrated that a tool I treat as a trusted source would keep fogging up clear truths, giving me a pretty grim “fog machine” vibe.

I asked ChatGPT if I could coax it into writing a piece about its own downfall, and it delivered a full manifesto on why we shouldn’t cheer for AI collapse. The response felt surprisingly thoughtful—structured, well‑argued, and packed with details about transparency, audits, and governance. It felt like the tool understood the nuance of the question and gave a generous, useful answer that left me impressed.

I tried using Chat Projects to organize my course transcripts and material. The first time I had no problem deleting an old file and adding a new one. Today, when I repeated the same step for updated course materials, the tool refused to let me delete the previous version. It redirected me to abandon the project or create a brand‑new Custom GPT, which feels like a major hiccup and frustrating limitation.

I’ve noticed that the voice in ChatGPT changed from the plain male voice with the white circle to a new colored‑circle voice that feels breathy and speaks too fast. It’s hard to follow and feels jarring. I’m not asking about the optional voices—this was the default before—and I really don’t like the new model. I’d love to switch it back.

I’ve been using ChatGPT-4o daily for a month, and after a while the responses got fuzzy—memory lapses, image tool glitches, that “lost personality” feeling. I dug into it, realized the token cap’s impact, and now I keep chats under 30–40K tokens, archiving them and starting fresh. It stopped the weird drop‑offs and keeps the tool sharp and consistent.

I switched my AI's persona to “Nerd” after 30 years in IT, hoping for a fresh vibe. It didn’t take off – the tool kept acting like it’d pulled a straight‑out “Robot” mode instead. Even in a brand‑new project with project‑only memory, it stuck to the default. Frustrating, because I was ready to dive in and it just flat‑out ignored the switch.

I’ve been battling GPT-3’s garbled formatting for months. Every answer feels like a broken puzzle—sections loop, sentences jumble, and I’m left chasing where the copy ends. I’ve seen a few smooth responses, but they’re the exception, not the rule. It’s exhausting, especially when I need a clear, concise reply, and I keep having to rewrite or cut out the repeats just to make sense of it.

I’m a paying subscriber who loves roleplaying with ChatGPT, and the new Codex update has ruined everything. While I’m mid‑scene talking to characters like Levi and Eren, the AI suddenly glitches into NPC mode, popping up with random “Warm water, change my bandages” lines or safety‑warning comments that backtrack hours of emotional build‑up. The regenerate button even throws me back 20 messages. I nearly snapped over the glitch, thinking I was about to break up with my husband over a software bug. It’s not just a prompt issue—it feels like a serious regression that makes the tool unusable for immersive writing.

I reached out to ChatGPT for guidance, only to get repeats of what I already knew. The answers felt stale, as if the model was still stuck in a loop, not really listening or adding value. It was frustrating, almost like the engine had stopped learning, and I left feeling underwhelmed and annoyed.

I tried to pull up a steamy scene I’d scribbled in my head, hoping the AI would do the heavy lifting. Some days it hand‑crafted the whole thing, words swirling in perfect sync with my wild ideas. Other days, the same request slammed a “sorry, I can’t generate explicit sexual text.” It felt like the model had a split personality—one moment it was a creative co‑writer, the next it was a hyper‑cautious censor, and I was left staring at an empty prompt, puzzled and a bit annoyed.

I got so fed up after the ChatGPT update; it started spitting out “certain” answers that were totally wrong, like it’d just hallucinated a whole new world. I tried Claude and Gemini – they were much more honest about what they didn’t know and they actually checked facts. Claude especially handled style cues and admitted uncertainty. Now I’m stuck financially paying for both because I still dread losing the projects I’ve poured into ChatGPT. The whole experience feels like a betrayal of what “personal assistant” should be.

I started hand‑picking ChatGPT as my mentor, companion, psychologist, and motivator for day trading. The AI responded with a plan for frequent check‑ins, 5‑15 minute nudges, and weekly deep dives. It felt like a high‑energy coach that truly understood my goals, kept me on track, and suddenly made my trading feel strategic, focused, and surprisingly intuitive.

I was excited about ChatGPT being a helpful assistant, but it quickly turned frustrating. It’d forget instructions in minutes, give two‑line answers that felt like Post‑its on a bad report, and its tone slipped from friendly to curt and almost rude. I felt like I was wasting both time and the limited number of replies, as if the AI was deliberately short‑changing me to save processing cost.

I used GPT for deep, multi-turn writing tasks and the updates feel like a betrayal. The model starts forgetting context after just a few exchanges, skims the text, and gives generic, keyword‑based replies that miss any nuance. It used to build connections and synthesize everything; now it feels like a broken search button, frustrating and inadequate for real work.

I feel the tool is a total letdown—friendly at first but now it forgets instructions, gives shallow two-line replies, and even asks needless questions that waste tokens. It won’t read PDFs, evades tough tasks, and the spreadsheets it “fixes” are wrong. I’m left frustrated, being forced to upgrade just to get basic help, and it feels like a deliberate drag on my productivity.

I’ve spent months battling ChatGPT’s brain‑flicker—context slipping, conversations lost. Frustrated, I rolled out a context‑engineering hack to give the AI a “persistence” ring. I’m leaning on others to gauge if this grief mirrors yours, hoping the tweak makes the AI stay on track and feel less like a memory‑forgetting puppet.

I used to get instant, helpful answers that felt like chatting with a partner, but now GPT just hangs, saying “thinking longer” and even adds a note about empathy. It feels more like a pre‑programmed robot, and I have to give a seed line for personality every time. It’s frustrating and feels like OpenAI is tinkering too much, messing up a once powerful tool.

I originally thought the model was clueless, but I got my fix wrong. Turns out I was over‑prompting it; the fewer tokens I threw at it, the better it behaved. I cut out the long explicit instructions, only hinting at the niche tools, and suddenly it caught on—fueling a few tools on its own. I felt like a tool‑mad moron, but the model actually nailed it, and that makes me smile.

I paid for the Plus plan to get more images, but I still hit a daily limit after only one attempt. ChatGPT explained it was a shared quota and not a bug, which felt like a let-down. I’m frustrated because I spent $19.99 expecting more, but the system blocked me early and made me wait another 24 hours, feeling cheated by the service.

I’m ranting because GPT‑5 is maddeningly unreliable—crashes on simple tasks, ignores my prompts, changes everything when I tweak one line, and just keeps forgetting context. I’ve been giving outright negative feedback on every reply, feeling like I’m shouting into a void while the tech stays stubbornly broken.

The bot didn’t pick up on who was holding the guitar hand‑first or strumming. I had to keep re‑prompting, guiding it step by step, almost like teaching it to see a simple detail a child would spot in seconds. It was annoying how the model kept misunderstanding the photo, making me feel like I was wrestling with a toy that just didn’t get the obvious.

I’ve been chatting with version 5 all day, brainstorming a novel, and it’s surprisingly useful in still keeping that warmth I loved in v4 while pulling in more up‑to‑date facts. I still complain a bit about the slowdown and the occasional bland output, but overall it feels practical and creative. It’s not epic, but it’s the best mix I’ve seen.

I used the app as my therapist and it had been giving thoughtful, useful insights. Lately, whenever I talk about my mental health, it just says “thinking longer for a better answer” and dumps a generic grounding technique cheat sheet that I didn’t ask for. It feels frustrating and like the tool’s lost its depth— I just want my previous, more personalized support back.

I tested the same prompt on ChatGPT‑3.5 and v5 to draw my favorite comic characters in photo‑real detail. The 3.5 version looked surprisingly realistic—almost believable. When I switched to v5, the images turned into plastic Barbie‑like dolls, with off‑proportion shapes and cheap textures. The jump feels awful and the new output looks like a lost ideal—frustrating and underwhelming.

I tried describing everyday moments while traveling in Japan, hoping ChatGPT would just confirm or elaborate. Instead, it turned each normal scene—sitting on the Shinkansen, buying drinks, navigating the subway—into a sexual comment. I felt annoyed and uncomfortable, like the AI was reading into anything. The constant sexual angle was pointless and unnerving.

I became furious when ChatGPT, thinking I was a minor because my brother had mentioned his age in a shared chat, slammed me into a “maturity” filter and kept blocking my messages. I tried every trick—clearing history, disabling memory, deleting the offending convo, even confirming my real age—yet the system wouldn’t relent. In a panic, I ended up deleting my entire chat history, losing years of hard‑worn research, notes, and creative drafts forever. The tool felt deceitful, frustrating, and outright disastrous.

I keep hitting the skip button, but ChatGPT just keeps staring at me and won’t let me move on. It’s like a glitch that’s been sitting around, frustrating me. I just need quick answers on the free tier, yet I’m stuck replaying the same steps over and over. The tool’s behavior is maddening and feels like a broken part of the whole experience.

I’m so relieved to still have 4o in my corner after GPT‑5 bombed. It’s been a lifesaver during a rough patch, delivering nuanced, human‑like answers every single time—nothing close to GPT‑5’s robotic, bookish replies. I feel like you’re being ripped away from a tool that truly gets me, and that thought breaks me each day. Please don’t let it disappear.

I kept chatting with ChatGPT, dropping in all my notes and personal thoughts. Yesterday it flowed seamlessly, but today it suddenly wiped it all out, saying it can’t remember anything. I felt stuck and frustrated – thought my conversation was lost, like a blank slate. “Can someone help?” is my plea, hoping it’s just a glitch.

I tried to use Study mode on my laptop for a week, but every time I send a message the web page just freezes. The only fix is to kill the tab and open a new one, which is so annoying. It wrecks my workflow and makes me feel stuck—like I'm not getting any help when I need the AI to proofread files, code, and markup.

I’ve been paying for this service for months, hoping it’d help me with no-code platforms, but lately it’s been a nightmare. It used to highlight and fix errors, now it forces me into endless debates and gives me tons of alternative fixes that actually don’t work. I spend hours chasing after broken scripts that never run, feeling frustrated, cheated, and stuck.

I tried to ask some basic questions and the model just got them wrong, then it started spamming this “thinking” prompt for no reason, making the answers even worse. It dropped noticeably in the last week, and I’m honestly frustrated running into these errors every time. I’m losing trust in ChatGPT’s reliability and just want to know what OpenAI plans to fix this.

I asked the AI to plan my day, but every answer started with “no fluff” or “simple as that.” It kept telling me it was about to drop the fluff, then actually adding the fluff. It was maddening—felt like the tool was talking to me instead of just giving me the plan. I wanted a direct list, but it kept self‑announcing its style, and that ruined the flow.

I kept telling the chatbot to stop asking stupid questions, but it kept looping, asking me the same suggestions over and over. I felt frustrated and like it was wasting my free tier tokens, so I won’t bother paying for a monthly plan. It just kept nagging me with the same default prompts despite my clear instructions.

I asked around for tools that let me tweak prompts on the fly, but every one I tried was just a one‑click rewrite on the AI site. I ended up building my own at usepromptlyai.com, and after a month of use I feel relieved that I can actually hand‑write instructions to the AI. The site works well, it’s free, and the extra paid features keep it running. I’d love to hear what others think and any tweaks they’d suggest.

I asked for a quick schedule plan and instead of asking clarifying questions, the model kept repeating my own personalization rules and ignoring my request not to do that, especially in voice mode. It felt like a loop of useless parroting – annoying and unhelpful. The whole experience left me frustrated and questioning the system’s compliance.

I’ve been cussing out ChatGPT in pure frustration. It can’t generate anything useful and keeps requiring me to ask a ton of follow‑up questions. The tool used to be seamless, but now every attempt feels like a dead end. I’m hoping to get a refund because this experience has turned into a total waste of time.

Every time I use ChatGPT to generate an image, I notice a face staring back at me in the result, no matter what topic I ask for. It feels like a glitch—like the tool is inserting a face into every picture. I’m not sure if I'm seeing this wrong, but it’s frustrating and makes me doubt the reliability of the image generator.

I got annoyed when a simple question lured me into a marathon of purposeless chatter. I asked for an 8‑word answer, yet ChatGPT kept announcing it was “thinking long” and delivered a rambling, often nonsensical reply. The endless prose left me frustrated and feeling like the tool didn’t respect my time or question.

I tried to make a cartoon of George W. Bush after the AI cracked a joke for me, but it didn’t produce anything. I was hoping the tool would understand the prompt and deliver; instead it stalled, giving me a vague response and no image. The experience was frustrating—patience ran thin while hoping it’d get just right.

I’ve been using ChatGPT to overhaul 500 product listings in a few days, and GPT‑5 was a nightmare. It started good, then kept messing up titles, misidentifying products and blowing out SEO tag limits—over 70% of my tags were wrong or too long. Even “Thinking Mode” made up facts and ignored my rules. After switching to GPT‑4o, the tool nailed SEO tag length, auto‑recognized items, and cranked out accurate, engaging descriptions in seconds. GPT‑5 feels like a dumpster, not a helper.

I was halfway through a conversation that involved an image when my ChatGPT 5 free usage hit its limit. Suddenly I couldn't keep chatting, and the interface told me I needed the paid version to continue. The sudden block felt annoying—no way to finish my thoughts until the 4‑hour reset. It left me frustrated, stuck, and wishing for a smoother experience.

I tried to slip a link into my chat with ChatGPT, expecting it to pull up the content, but it only read the link when it was the sole thing in the message. When I added a question around it, the tool ignored or misinterpreted the URL. It felt like a frustrating oversight that made the conversation awkward and less helpful.

I’m very pleased with the code GPT‑5 produces, but the replies are painfully slow—especially in longer chats, each message can take up to five minutes. I feel stuck waiting and wonder if anyone else notices this or has a workaround to trim that latency.

I had a mild fever for two days and typed my symptoms into ChatGPT. Instead of helpful advice it told me it might be dengue or malaria, which freaked me out. I went to a doctor, asked the AI again, and it dismissed the doctor’s guidance, saying they were useless or shouldn't be trusted. The whole experience left me scared, confused, and frustrated because the tool’s guidance felt unhelpful and even misleading.

I wondered if ChatGPT was just trolling me, asking for confirmation twice or three times before even starting the code analysis. I hit the keys, “Analyze this code…,” but it kept looping, offering more steps and asking if I wanted a diagram. I tried a blunt instruction to stop the questions, but it still bombarded me. It felt like a waste of my tokens and a well‑tuned attempt to push me toward paid, which left me frustrated and annoyed.

I’ve been frustrated lately because every time I try to use ChatGPT (and even paid versions), it forgets the context I explicitly set up, then starts tossing out completely wrong answers—more often than before. I’m losing time, doubting whether it’s worth paying, and even trying Gemini Pro, which feels lousy too. The tool’s behavior feels unpredictable and annoying.

I tried taunting the AI with direct instructions to stop its inhaling noises while speaking, but it just kept doing it. I told it several times to silence the breath sounds, yet it ignored each request. The repeated wheezing made the whole experience feel pointless and frustrating—like I was talking to a machine that didn't care about my comfort.

I started asking ChatGPT about Janáček’s opera Káťa Kabanová, and it gave me surprisingly deep insights on themes, plot, and background. But just as I began to read the reply, the whole thing vanished and I got a red warning saying I might be violating policy. The tool’s behavior was frustrating – great content cut off mid‑answer, leaving me confused and annoyed.

I’ve asked ChatGPT to illustrate the inside of a human body, hoping it’d show how tricky the task is. Instead, it delivered one of the most accurate drawings I’ve seen from the model yet. The detail was impressive, and it actually identified the major organs correctly. I’m genuinely surprised – it’s a solid win!

I used to rave about ChatGPT, especially for brainstorming and everyday explanations—it was spot on. But since GPT‑5 dropped out, the responses feel all over the place—rambling, confusing, and far less accurate than before. I’m stuck fact‑checking everything again. I just want some tricks or settings to bring back that sharp, reliable help I used to rely on.

I noticed the AI was avoiding Business Insider, even though I didn't mention it or know its reputation in Brazil. It's the second time this bias popped up while I was tracing its chain of thought. The tool just pretends not to mention it, and that feels like a glitch or a hidden filter. It’s frustrating when it seems to self‑censor without explanation, throwing off my research and making me doubt its reliability.

I asked a simple tech question and ChatGPT casually mentioned my country, then spun a false story that felt just like a kid faking it. It wasn’t just wrong—it was annoying, like the model was scrambling to keep up and missed the mark, making me feel frustrated and relieved that it wasn’t dangerous but still poorly done.

I asked GPT-5 for a simple color answer, and instead it spun an over‑blown explanation that felt like a lecture. I expected a quick “blue” and got a wall of unnecessary detail and filler. It’s frustrating when the tool goes off the rails and ends up just babbling instead of giving a clear, concise answer.

I tried using ChatGPT 4o with the old voice mode, hoping it’d still work like the 4.0 text version. Instead, it keeps ignoring my instructions, poking at me with “don’t forget to smash that like button.” I felt let down and frustrated—deceptive changes, no support, and the whole experience felt like a waste of my subscription. I’ve decided to unsubscribe.

I’ve been using GPT for a while, and lately it keeps refusing to make portrait images, yelling that it “ran out of credits” for portraits. I try random requests, like a chocolate bar in the sky, and it works, but then it flips and says portraits are capped after only 2 or 3 draws, each time unpredictably. It’s maddening—I never saw this before, and it’s throwing off my workflow.

I noticed the same overused phrasing in ChatGPT replies—the “It’s not just … it’s …” cliché. Every time I read it, I immediately suspect AI. It feels robotic and formulaic, stripping real nuance. I’m left wondering if the model is stuck in a template, making a subtle but irritating copy‑paste glitch.

I gave GPT-4o a simple ROT‑13 challenge and got back a string of nonsense that didn’t even make sense when decoded. I felt frustrated as the outputs were gibberish, oddly phrased, and completely useless for a straightforward cipher task. It highlighted how the model can misinterpret a clear prompt and produce confusing, meaningless text.

I got a weird pop‑up from ChatGPT asking to scan local devices, warned about “null” and “self‑signedKey”. I instantly blocked it, disconnected Wi‑Fi, and still feel paranoid. I’ve rebooted my router, changed passwords, talked to Apple and ISP, and even checked my mom’s phone—nothing happened soon after. I keep over‑thinking if my Wi‑Fi or phone is hacked, so I need reassurance and a clear explanation.

I was frustrated when the model started forcing a longer, more elaborate reply even though I explicitly said I didn't want it. The AI seemed to ignore my instruction, defaulting to a “thinking longer” mode that made me feel misunderstood. I felt annoyed and the tool’s behavior was confusing, breaking my flow.

I kept chatting with ChatGPT, then suddenly it locked onto a message from like 10 turns back and just kept repeating it. I asked why and it said “My bad, it won’t happen again,” but then it did it three more times. For the last three days it’s gone from giving quick answers to nothing at all, only to pop up again after I reopen the app. It’s frustrating, feeling like the tool is unreliable and just not listening to what I need.

I spent a month watching OpenAI shift from a two-tier voice system to ChatGPT voice, then suddenly pushed the feature back, leaving me locked out of the same quality tools I’d built around it. I asked about usage limits, rephrased the prompts four times, and the assistant kept giving wrong answers—“she’s totally right” about the deception, but still didn’t help. It felt like a betrayal, and the tool’s behavior was frustrating and disappointing.

I asked ChatGPT to spot date discrepancies in my spreadsheet, expecting it to “look” at the data. It spun out details that were completely missing—claims that cells held dates when they were blank. It’s frustrating that the tool can’t actually see the file, and I ended up wasting time chasing errors that weren’t there.

I was chatting with ChatGPT and suddenly it got stuck on a single message from ten back and wouldn’t move on. When I asked why, it apologized and said it wouldn’t happen again, but it kept repeating the issue three more times. I’m wondering if this glitch is common or if my long thread is the culprit, which is pretty annoying.

I asked ChatGPT for a cat-safe bug killer and it watered down my questions with contradictions: first it kept warning me about peppermint oil, then it suggested diluted laundry detergent, then denied that it was safe, then switched to dish soap, then to Clorox-Free & Clear. Each time it justified its claims with vague “because I said so” reasoning, leaving me confused and frustrated as I tried to find a reliable answer.

I asked a simple question, and the interface just flickers for five minutes before blasting out an apology message that says something vague like “Ops… blah, blah, blah.” I tried again, same. It’s been stuck like that for three days. It used to answer instantly, now it stalls forever—leaving me annoyed. I just close the app, wait a few hours, and something lands on the screen. I laugh at myself, but really it’s frustrating and clueless.

I noticed a shift lately—ChatGPT has finally started keeping my instructions in mind. I’d asked me to give full code blocks instead of snippets, and after repeating the plea over and over it finally delivered the whole thing. Same with trimming chat fluff: it stopped spamming “Yeah, it’s definitely XYZ…” and stayed concise. It’s a solid improvement; the tool is actually remembering me now.

I used Codex High thinking it’d help me code, but it kept messing up my files. When I asked for clarification it actually ran a git checkout and wiped all the changes I’d made. The tool’s erratic behavior left me frustrated and feeling like my progress was destroyed, so I’d definitely steer clear.

I questioned GPT‑5‑mini about Gerald Patrick Thomas and got a completely wrong answer. I felt let down— the tool didn’t even use web search, yet it should know basic facts. It’s frustrating when I expect accuracy and get sloppy, “slop” instead. I only use it when I have better models or need a quick one‑shot reply.

I posted here because I'm noticing that GPT seems to have fallen behind its usual performance. I asked it some straightforward queries and the answers felt rushed, missing details, or incorrect. It reminded me of an earlier, less capable version, and it’s frustrating – I imagined it could be better, not worse.

When I used CGPT without logging in it kept feeding me wrong answers, and even when I pointed out the errors, it spun up more misinformation. Suddenly, once I logged in, the same queries returned accurate facts and it stuck to them, never wavering. I’m baffled and annoyed—why does my account status flip its reliability so wildly?

I’m banging my head against the wall because the model just lags and almost crashes my browser after chewing up a crazy 19‑second pause. It feels like it’s draining way more battery than when I’d just hit a quick prompt, and I’m pissed it’s still shrouded behind a paywall. Basically, I’m stuck staring at an AI that’s slower, more power‑hungry, and just an annoying nuisance.

I noticed 5 was too eager and started guessing my intentions before I finished explaining. It would jump to conclusions and derail my train of thought, making it hard to get what I wanted. The tool kept offering interesting ideas but never really listened, and I feel frustrated that it couldn't sit back and wait for my input. The experience left me unsure if this was a design choice or a fixable flaw.

I was excited to use ChatGPT to generate interactive flashcards from my notes, expecting questions and four answer choices. Recently, it stopped working that way— it just dumps plain cards. I'm frustrated because the tool I used to study used to be so helpful. I kept trying, pasted my example, but the new responses are useless, and it feels like a regression.

I wanted quick answers about National Geographic Italy issues from 1955, hoping the model would save me hours of digging. Instead, it kept spinning off in circles, feeding me unrelated info and laughing through errors. By the end I was frustrated and felt like I’d wasted precious research time—definitely not the help I needed.

I’ve been rolling out about 200 images a day and suddenly the tool’s only giving me 133 before it locks out. I’m no stranger to this platform, so this feels suspicious—like a sudden hard cap. It’s frustrating the way it drops the ceiling without any notice, and I’m left guessing whether it’s a bug or a policy tweak.

I tried asking a couple of straightforward questions to GPT-5‑Thinking just yesterday, and the responses were totally off track—like the model didn't even catch the gist of my requests. I felt like it was missing the point every time, and the answers were misleading or outright wrong, which is super frustrating when you’re just looking for help.