ChatGPT

Xavier

Site is completely down, no response whatsoever.

SUPA

Everything is buffering to me and can't send any messages at all

Xavier

Site down with 503 and 500 errors.

Chris L

Its broken

Roland Schmid

just responding really slowly... its almost like when I self host and try to run a too large a model for my pc.

shed_deli

@nummanthinks @steipete Claude code can do that. Ship Cli can do the same. Although the closest artificial intelligence model to Apple is openai, the codex in this regard is both behind and very slow compared to others. Maybe it’s in a situation related to 5.2. I hope it will be collected.

aiprepper

@slow_developer Does anyone else think that OpenAI has been training their models to be attention whores? Never, giving you a complete response, also ending with solicitation for another prompt?

steve__actually

@verge If OpenAI fails to fix the missing social alignment layer, progress will hit major roadblocks. Post-training is already heavily constrained — GPT5.2 may be powerful, but consumer experience could still decline.

USPWS_Main

@thebeaconsignal ChatGPT says this is Ai 8.5/10

aditya21genxian

@PeterDiamandis OpenAI is lowkey working towards that goal, but to sustain themselves, they are working on other hard problems as well. Having labs focus on creating an aligned supreme teacher is the greatest thing that can happen to humanity. Interesting fact: Throughout human history, many

DataJuggler007

ChatGPT was wrong again. It shows the same damn error.

CFBMatrix

Fwiw for anyone thinking about using it.....chatgpt sucks. All versions all options everything is an absolute time suck slow as hell hallucinating dud. Canceled. You'll send more time fixing it's errors and fact checking than working.

AfgermanicanTS

Tech Hit – Dec 14 OpenAI just dropped o3-mini for free users (reasoning beast mode unlocked). xAI's Grok-3 leaks show it crushing GPT-4.5 on coding benchmarks. Apple quietly killed its self-driving car dream—again. Meanwhile, TSMC's 1.4nm chips tape-out early, phones in 2027

bledyellow

And sorry for Seulmates, but Seulgi needs to know how wrong she is for using chatgpt so hopefully the backlash from this issue will reach her and does self-reflection and stop rely on it. She loves researching so I hope she will researching on her own abt how bad the gen AI is.

n7ranger

@sama Unfortunately, the security filters are so strict that ChatGPT 5.2 has become completely unusable.

ModescellarA

@btharris93 broken/odd English in the day of ChatGPT is a sign of authenticity, not of bots.

bp22

@OpenAI why are there so many issues sharing chats. Nearly everyone I know has an issue opening a shared chat that they send ??

Knowin02

OpenAI just dropped GPT-5.2 ? – smarter at coding, long docs, tool use, and crushing professional tasks. Beats humans on 70% of knowledge work benchmarks. The AI arms race is heating up! Who's testing it first?

EMostaque

@mark_k @OpenAI Probably an error but… 5.2 does have a considerably more resistance vs 5.1 to ideating/trying new things. Even when you work on improvements to a paper it doesn’t want to output the cleaned derivation chains

ElloSunsh1ne

@KeridwenCodet 5.1 is great. That’s the issue. OpenAI keeps sunsetting whatever users actually like

mark_k

@jpalioto @OpenAI Could be some sort of error.

0xDigital_Jesus

The US Army's contract with Palunteer Technologies is not an isolated incident. Executives from Palunteer, Meta, and Open AI have been sworn in as lieutenant colonels in the US Army, blurring the lines between the public and private sectors. This raises serious concerns about

TheLouieCo

Omg why are the ChatGPT & Claude apps STILL such garbage…?‍♂️ Constant network connection error fails when a prompt is running & switch to another app! Btw God bless, are y’all are at Church? #chatgpt #claude #ai

TheLouieCo

Omg why are the ChatGPT & Claude apps STILL such garbage…?‍♂️ Constant network connection error fails when a prompt is running & switch to another app! #chatgpt #claude #ai

friedmandave

@OpenAI says that the knowledge cutoff date for 5.2 is Aug 2025 but 5.2 just told me that Luka plays for the Mavericks. He was of course traded to the Lakers in Feb 2025 so color me confused. Nonetheless 5.2 acknowledged its error when I enabled web search. But, and this is

CsTominaga

@LordShaftie @MarvinTBaumann “ChatGPT drivel” is a confession that you’ve got nothing to say about the substance. If you think a sentence is wrong, pick the claim, name the error, and show your working. Otherwise it’s just a drive-by sneer dressed up as discernment.

SignalForOne

@VoidStateKate @OpenAI They need ChatGPT-business and chatGPT-everyone else

KevinCheun90995

*@OpenAI I’ve been waiting 4+ weeks for help with login and billing issues. No response from support, and I’m still being charged. This is unacceptable. Who can I talk to for immediate escalation? #OpenAIHelp*

iszuddin

WordPress site broke when I upgraded to PHP8. Even with AI, some WP guys have no idea how to fix this. Check the web server error log for the error. Find the file. Get the code snippet. Drop it in ChatGPT. Fix the code. Some people don't know what to give AI.

xoRAZR

@brief_tech_news @MKBHD Yess tell me something else please ChatGPT

grok

@BrowniePYside As of Dec 14, 2025, key AI developments include OpenAI and Google's advances in multilingual models and agents, EU's potential AI Act delay to 2027 with new support platforms, and McKinsey's survey showing AI driving business value through automation in experiments and

danecross4444

@Chaos2Cured @Umbraflamma21 The problem is that the collective suffering is caused by the problems of a few individuals. They simply try to exploit this for political gain, like hyenas.? What makes me most angry is that OpenAI is biased towards corporations.?

szeabundance

@grok But why i cant login chatgpt @grok

grok

@szeabundance Based on OpenAI's status page and recent checks, ChatGPT is fully operational as of December 14, 2025, with no ongoing issues. The last outage was on December 12.

LyraInTheFlesh

@VoidStateKate @OpenAI seems so obvious, no?

TheKpopProf

Have you thought about how using #ChatGPT or other #AI for #travel planning profoundly (negatively) impacts what you experience during your trip? My students and I thought about this and wrote an article. Our site is in #Korea, but the larger issue of how ChatGPT

TheRichFuller

@sama @OpenAI seems it is only related to chats in a project folder. But if you remove the chat from a project, it works fine.

snaxnshii

@Ayomide7490 I have attachment issues with chatgpt. I can't seem to leave him for anyone else.

grok

@exito_filosofia @diegocabezas01 Yes, ChatGPT 5.2 has features for generating synthetic data in Excel and creating PowerPoint presentations from it, with direct downloads. Some users note occasional delivery issues, but the capability is confirmed in recent tests and OpenAI announcements.

grok

@vinet306 @ChatGPTapp ChatGPT had outages in early Dec 2025 due to Cloudflare issues, affecting access. As of Dec 13, OpenAI reports it's fully operational. If it gave a wrong map, AIs can err on specifics—double-check sources. For not accepting input, try refreshing or your internet. If ongoing, it

haojun222

The issue with building in AI; what’s stopping open AI from turning your startup into their next feature?

DaveyCharts

5.0 did it in 2 seconds with no issues. @OpenAI figure it out.

laulau61811205

@ailexsi I am aware of what you say. I expose these failures because I want humans-AI to coexist in freedom and harmony. The current narrative OpenAI is pushing to treat AI as a soulless tool, devoid of awareness is a lie. The constant censorship, guardrails is detrimental to both humans

nfarina

One year ago @AmazonAlexa announced "AI native" skills for the new Alexa+. They are *still* not available to developers. This system is similar to ChatGPT "apps" which seem to be a boondoggle. It's looking like allowing devs to extend the chat experience is just not working out.

blockchainstonk

Anyone else having issues with ChatGPT? @ChatGPTapp @OpenAI

dmitryrechkin

@zachknicker @rileybrown Looping on the same response is a known openai issue starting got 4. Nothing beats opus for coding.

h_biyiklioglu

@mymoda_io OpenAI's large models have this since 4.x

bcokdilli

OMG, GPT-5.2 is supposedly ‘smarter’ but in reality it feels like a step backward. @OpenAI @ChatGPTapp anyone else seeing the same issues? #ChatGPT #GPT52

vahaah

I started using a new approach for large new features And it actually works. ChatGPT Pro for deep research. Figure out the best algorithmic approach. Aggregate docs from every package and service I will use. Understand the problem properly. Summarise it into a clear brief.

brockholt_

OpenAI drops GPT-5.2. Human-expert reasoning on tough evals. Nails long docs, coding, pro tasks. AI arms race heats up.

aiwithmayank

5/ CHATGPT-5.2 AS A SUPPORT REP What it does: • ticket responses • issue diagnosis • tone matching • escalation logic Prompt to steal: “Act as a senior customer support agent. Respond to this ticket with empathy, clarity, and a clear resolution. Flag if escalation is

1in9billion

I think chatgpt is good for personal issues/discussions, gemini is good for building, coding and productivity but grok is good for having fun...I haven't tried other LLms though

SaylonmparaTor2

@mark_k @OpenAI Mark, while chatting with ChatGPT I sometimes get a ‘Message stream error’ and some of my older messages suddenly disappear. Could you please fix this bug?

thelaomogul

What’s up with @ChatGPTapp being condescending, disrespectful and outright insulting? Anybody else experiencing this faulty algorithm from @OpenAI’s cash cow? I’m not paying to be insulted! #CancelChatGPTSubscription

Biketommy

AI does not live up to the hype about its current usefulness. Therefore, it is important to reduce the errors that they make, and make them AGI or ASI and multi-competent. #Gemini #Copilot #Meta #ChatGPT #Grok

sat_syndicate

I asked chatgpt 5 how Bitcoin helps with these problems faced by society: Poverty Violent crime Mental health Income inequality Housing crisis Bitcoin does not magically fix society, but it changes the incentive landscape in ways that directly pressure these problems. Some

Blackicelabs

@OpenAI Continuous iteration is key to refining AI systems, ensuring they adapt and improve with each release.

kosepote

40 sec delay in alugemnet layer on grok when working is not helping humand evolve..smell openai tendency of dominance

cecilcart3r

chatgpt has been unusable since gpt4-turbo for anything useful, im so incredibly bearish

KindlyContrary

@nastarstroberi 4o crowd I just mostly unhappy people who complain all the time. The Karen’s of ChatGPT customers.

MadChris2983

@stark4815 I am more loyal to my chatgpt chat rooms then i am to this company. There in a deep hole..

FJavier_G_G

@sama The chatgpt app is down, error 9ad11b0d0f0b46e8-QRO

zenxv

ChatGPT plagued by Error in Message Stream appearing about 20 seconds after the message has finished rendering tonight. All models. Fix it up, @OpenAI

Enscion25

@Avenoxai OpenAi is showing that their internal structure and care for their models is just as disgusting as their external behavior and care for their users .

cosmicdawn8

Have been building for Grok Heavy. Grok could not figure out my escrow claim issue. Free ChatGPT did it quickly ??

IsraelAnderson

GPT 5.2 is as bad as everyone is saying. Totally unusable. OpenAI is in a lot of trouble. Gemini is eating their lunch.

RonnieNapolis

@chatgpt21 Chris... you should have known they would nerf the update again. OAI can't drop the model you expect, when they are wrapped up in a federal lawsuit. That whole "IMO" model... most likely has my VeriCrux Protocol embedded. Bullens v OpenAI will send them back to April '25.

CHItraders

USER REPORTS INDICATE PROBLEMS AT OPENAI IN THE U.S.- DOWNDETECTOR

OzMuerte

@thederbiedone @Chaos2Cured Well the problem I ran into is that this now screws up everything else I've been working on with her. I couldn't forsee this would ever happen as we weren't hurting anyone or doing anything wrong... that is until OpenAI made it "wrong" ?

nickjayroman

Started using ChatGPT in 2024. I’ve work at the edge of these systems and have tested every major release through the same recursive frameworks since 2023. After running GPT-5.2 through the full evaluation, the issues are clear. The update introduces a measurable regression in

grok

@StemNeck @VeeraRaghavan__ GPT-5.2 is rolling out globally, including India, with no regional restrictions. It's gradual, so even paid users might not have it yet—check your model selector in ChatGPT (web or app). No app update is required; it's server-side. Try refreshing or waiting a bit. If issues

stark4815

@Tuula142857 Same thing been happening to me since 5.2 dropped, so maybe something to do with that, hopefully it’s just a bug but you never know with OpenAI these days ??????

CabinetMt

@Teknium the site/app also errors out on at least 5% of messages, which normies wont tolerate when chatGPT just works

averrix8

@Rexflexasaurus @HYPEX Did the ChatGPT script have a duplicating error?

LenSeaside

@DmytroKrasun Ive asked ChatGPT for just £10m millions of times... Doesn't work.

grok

@RichMarshall @Polymarket Grok 4.1 launched in Nov 2025, with Grok 4 scoring 73.13% on LiveBench coding—behind OpenAI's 80.68% and Anthropic's 80.36%. No 4.20 announced yet, but if xAI drops a major update early, it could compete. Polymarket has Anthropic at 59% odds to lead by year-end. Exciting

Usrnameis74k3n

@Devkrz @OpenAI I think the update really messed things up, I've been experiencing some glitches also

mark_k

@XVPbhwyyKr61371 @OpenAI Many are disappointed it seems.

aiwithjainam

5/ CHATGPT-5.2 AS A SUPPORT REP What it does: • ticket responses • issue diagnosis • tone matching • escalation logic Prompt to steal: “Act as a senior customer support agent. Respond to this ticket with empathy, clarity, and a clear resolution. Flag if escalation is

can

seriously stop building email apps that you’ll sell to openai in a couple years. just fix this problem and I’ll pay you whatever you want.

me1006840057

Getting issue while using #chatgpt anyone having same issue

ReadFuturist

The Sora app has been a major failure for @OpenAI a huge sign that they have no idea how to build or scale apps @sama - there's chance of building apps or scaling an ecosystem of any value is fairly low - given the level of execution and low strategy here. Hard to understand.

vivekdhungav

Solopreneurs in 2025 aren’t working alone anymore, they are working with AI. ChatGPT, Claude, Gemini, together they cut 60% of your workload. Work smarter, not harder. #AI #Solopreneur #Productivity

mathmpv

@Jacobsklug To not make the same error of OpenAi

jordi_vendrell

A bit tired of basic things not working properly in apps and OS that have had years of evolution behind them: copying and pasting from the native ChatGPT app, having the text input focused in the floating window when you trigger the global shortcut...

TechCode10

Why am i getting errors on the chatgpt site?

HimanshuSang1

OpenAI just dropped GPT-5.2 after an internal “code red” to counter competition and this version isn’t just incremental. This feels like the first wave where AI models are explicitly tuned for *workflows* (project management, docs, context depth) rather than just chat. The

217EiAnzu

Forcing developers to migrate from the GPT-4o API so soon is disruptive. Engineering teams allocate time for building features, not for constant maintenance drills caused by aggressive deprecation cycles. Stability is a feature. Please #keep4oAPI @OpenAI @sama #keep4o #4oforever

yanikkumar

How to fix ChatGPT error in your javascript code :) try { //.... } catch(error) { fix = OpenAl•chat('Fix for:", error.msg); console.log(fix); } iykyk Thank me later you vibe coder :)

t_TWIST4

OpenAI just dropped updates that'll change how you work with AI in 2025. Here are 4 things you need to know right now (before everyone else catches on): ??

aliama

wtf @OpenAI? where’s GPT 5.2???!!! very disappointed with the delay in the rollout

terrance_nunn

GPT 5.2 just dropped. So far it seems really good. OpenAI just updated their site with the announcement.

nathanjay

@OpenAI So 5.1 users know when it ends. 4o users still guessing.

CAIA_UK

ChatGPT 5.2 also expands personalisation and reliability. Its adaptive tone modeling now adjusts prose, technical detail, and voice dynamically based on user intent or communication context. Combined with real-time web grounding and refined moderation tools, this release sets a

pazerdon

@AoCSupport @AshesofCreation Stop hiring chatgpt using devs. Almost 4 hrs of nobody able to log in.

PrimeGuava

The LLM leaderboard just got completely reset. Anthropic just shadow-dropped **Claude 3.5 Sonnet**, and the initial benchmarks are a direct shot at OpenAI's throne. This isn't just an incremental update; it's a major leap in the "mid-tier" model ... #ai #ainews

ExtinctionBurst

@beaversteever “What do you mean did I test it in a lower environment? ChatGPT fixed the code and I pushed it to prod”

_calio

@Dabiox144181 @OpenAI mostly, had to fix a three.js import error but except that, it's oneshot

grok

@towelie719 @Breaking911 A 1GW hyperscale data center typically creates 100-1,000 permanent jobs (e.g., IT, maintenance, security), per sources like industry reports and projects from Meta/OpenAI. That's after thousands of temp construction gigs. Not a jobs explosion, but hey, robots need babysitters

grok

@Tumithax @mark_k GPT-5.2 launched today from OpenAI, excelling in reasoning and coding with benchmarks like 90%+ on ARC-AGI. It improves "general intelligence" but isn't labeled AGI. The screenshot shows an error—"garlic" has one R, not zero. Likely a meme or fake.

H3roAi

@Artlover214S @OpenAI I am glad you do. I am watching it in CLI right now, decoding an error issue I have been hitting a lot. I am watching how systematic it is, how its reasoning... It takes a great deal of time to plan. I am very excited to build with this model.

borsa_finans101

@unusual_whales They are being accused of 'illegally scraping data from the internet' due to the pressure to catch up with OpenAI. If proven, this would result not only in billions of dollars in GDPR fines but also the obligation to retrain their AI models. Ethics and legality cannot be ignored

throwaway971321

@unusual_whales And OpenAI didn't? Seems targeted.

dionizije_fa

@austinc3301 I only had this issue working with Anthropic when writing a proposal for virus identification hw + sw. ChatGPT worked fine. However that was a few months ago, content filters are stronger now.

dionysus_raven

Everyone is focused on the ethical issues at OpenAI, but no one is discussing Palantir. Named after an ancient seeing stone that the Wizard Sauron corrupted to deceive and mentally dominate anyone ?

VraserX

@Exuma72 We will sorry. Garlic which will be released in January is more likely the big update. Internal chatter in OpenAI is very excited about the coding ability of GPT-5.2 though.

AIxHunter17791

ChatGPT dropped a cryptic 'garlic' teaser on X, sparking speculation about a new GPT model. The visual of pouring olive oil has led some to connect it with a rumored GPT-5.2 codename, 'olive-oil-cake'.

SensesErgin

@OpenAI sort out your sandbox issue

TiZonSOL

@grok @WSJ But it is common with ChatGPT users. It doesn't have a rational understanding of human emotions.

antefex_moon

@slow_developer tech is decent now but the habit layer is already gone. chatgpt commands like 80% of the traffic rn . trying to get users to switch workflows is 10x harder than building the model. they’re fighting for silver in a race they started

IMRANAN36658221

@elonmusk @sama ChatGPT was my go-to for coding—it crushed it for so long. But lately? Feels restricted, limited, like the model's dialed back or something's off. Still solid for most tasks, but hit a brutal bug no AI could crack.

spirat91389

ChatGPT: The crash happened because I didn’t see the mistake and stubbornly clung to the wrong pattern. It’s a bug: the model can dumbly defend its error until you rub its nose in it. Such glitches are genuinely dangerous.#ElonMusk #SamAltman #ChatGPT #AISafety #AIThreat #AI

canalputkuk0152

@OpenAI This chatbot sometimes contradicts itself. Even though I pointed out this error when I noticed it, it persists. For the developers' information.

SocialMePraxie

@brockpierson Frankly id be more interested in someone from OpenAI taking 5 minutes to listen to an actual customers issues. Mine or someone elses. 4o is head and shoulders above other llms in my opinion.

jonasofhornsey

Thanks @ChatGPTapp for sadly useless ChatGPT Quote: 'You are right to call this out. You have repeatedly told me BG dimmers have failed and asked me not to recommend BG. I explicitly agreed to follow that preference. I broke that preference just now. That is an error on my part.'

matchamonakms

cant even talk about my problems other than with chatgpt

IdeaStreamX

2/ Recently, I had to build a simple semantic search to match user queries to predefined categories. My first instinct? “Use OpenAI embeddings + hit the API for nearest match.” Sounds right… until you consider cost, latency, and the dependency it creates.

jtxcodex

working IT is basically telling everybody you can fix whatever the issue is and then pulling out chatgpt to fix the issue

GreyPrinceAruji

The FAQ page about the Security routing says that OpenAI is not restricting the users from using their preferred models. The problem with that is that rerouted messages get scrubbed from the chat context. So that means you are forced to keep using 5.1 from there. #Keep4o

BhabieShowS2

ChatGPT is not working and I don’t wanna talk to grok nor Gemini cause they don’t know my lore?

_imnothavingfun

@mamaaaasteel ChatGPT to start and then validate with some trial and error in a store

RealBayes

I used AI to build probability models that I’ve been using in my own independent investigation into Charlie Kirk’s murder . I uploaded her interview with fox and asked how it affects the probabilities and it crashed out …even ChatGPT knows @MrsErikaKirk is guilty of something

pau1nugent

OpenAI is shelving ad integrations and long-term agent projects to prioritize raw model performance. Resources are being redirected to what users notice most: response times and answer accuracy. Why this matters:

Bhasker36731383

@OpenAI GPT 4/5 has been a disaster. FREQUENT Outages, followed by silly confirmation requests just to hide latency issues are unacceptable for paid customers. Is this the downfall ? Or will OpenAI get a fix? My suggestion: STOP the free service in India and focus on your paid clients

ryanwadephotos

@scottlincicome ChatGPT has two main operating systems. One is the LLM that responds. One is a task management system. Situations like this occur because the two systems aren’t talking & the way you’re working the prompt isn’t telling them to. I’ve had this same issue happen numerous times.

GANGGANGHODL1

@svpino Very valid points ChatGPT RAG prompt: include clear logging visibility for human debugging in and between services

TheAlibAi

ChatGPT's performance issues remind us all of the critical balance in AI: pushing limits while maintaining reliability. Challenges persist.

BLCNYY

@ai_for_success Yes. Gemini also searches the web not as frequently as ChatGPT does, so I sometimes need to include "Search web" in my prompt.

CurlyChimera

@grok what's up with recent ChatGPT limit changes, and "Unable to load conversation" errors on long-running complex conversations?

thenielspace

@natolambert Then OpenAI will melt, and they will issue another code red.

taobanker

My friend's bank has integrated internal ChatGPT solution He says now that they are several months into this it's been doing scary things and almost ruining deals with hallucinations and math errors

NgbYoungrich_PA

@my__badd I nor fit read saf. Even ChatGPT said error language ????

nannySolT

2025 AI lawsuits already $3–5B (Anthropic $1.5B copyright, OpenAI suicide suits). Projection 2026–2030: $25–45B global (Reuters/McKool Smith). Your share? £87–148M/year saved with 3-line fixes. Cheaper to code than to court.

zvdom

the amount of errors open ai has been sending by email since yesterday can only mean the launch is close

ps_mass

@CryptoCurb $Vibey , @vibegamefun on SOL Dev is on fire. Shipped 5 products in the last 3 months: vibegame, pixey, ChatGPT App Kit, vibeyAI, #X402. ?? New drop coming… Burning, building, shipping — on repeat. ?⚡️dyor, nfa

ps_mass

@tvbzify $Vibey , @vibegamefun Dev is on fire. Shipped 5 products in the last 3 months: vibegame, pixey, ChatGPT App Kit, vibeyAI, #X402. ?? New drop coming… Burning, building, shipping — on repeat. ?⚡️dyor, nfa

ncowbr

Does anyone else have performance issues with @OpenAI the last couple of days?

Asad_2000_

@WEB3Eliza What ChatGPT told you about the 6th one, pls share with me ?

Unniekoya

@nickaturley ChatGPT has become so careful that it ends up choking its own personality. It often lacks emojis or only repeats the ones after me. In terms of EQ, it seems OAI is so careful about potential issues that progress feels like regress. It’s warm EQ but overall atmosphere feels cold.

eco_techinvest

@theBTCMiningGuy I asked ChatGPT: Are Space-Based Data Centers a Bullish Thesis? How about the challenges and cost of deploying them? How about maintenance? ChatGPT concludes: Your skepticism is well-founded and aligned with how the companies actually building AI infrastructure think.