Replies dated this day in the collected sample: 187. Collected since 5 Oct 2026: 349.
X replies are found by search and each one is checked against its public post on X. Last read: 26 Oct 2026, 23:46 UTC. Not every reply is read.
- A problem
It feels like Codex is going backwards, not forwards. It can't help edit code like it used to in Shopify. It's screenshot, copy paste again. Like it was 3 years ago. Back to Claude everytime.
@kaganpowell on X ↗ · 00:00 UTC
- A problem
Watermarking and full cutting of use and dumbing down models all in one go .. worst month for openai ever.
@AVAudioVisions on X ↗ · 00:01 UTC
- A problem
It can’t soft the problem for 5 hr limit for plus user
@lastrada888 on X ↗ · 00:11 UTC
- A problem
Day 2 of what? It doesn't even seem like you did anything. Nothing has changed. It's extremely slow. Usage runs out extremely quickly and I don't even have any desire to use a dot. .Dot was the right name, though, because it's like a speck, something you don't even notice.
@Wood_Unlimited on X ↗ · 00:12 UTC
- A problem
Downgraded my plan from pro, charged today for my new plan and yet I have no usage until October 9th. Any help here Tibo? Have only been able to use usage credits from my RepoPrompt bonus @pvncher
@mikeinprogress on X ↗ · 00:14 UTC
- A problem
WHERE IS MY RESET
@wuweiweiwu on X ↗ · 00:15 UTC
- Mixed
it was pretty weak today, but I would say reset tomorrow. Day 1 was good in theory but actually not that great, speeds on SOL 6.1 still slow.
@voicesintime on X ↗ · 00:19 UTC
- An idea
Btw. could usage limit resets that are expiring automatically be applied upon expiring? Or show the exact time of day when they are expiring when hovering over it. Thanks. I've lost two resets because I wanted to drain my remaining % first, but then forgot to use them in time.
@s3sebastian on X ↗ · 00:36 UTC
- A problem
Please take a good look at its Tokens output efficiency, which is now ridiculously slow. This is a continuous task of more than 30 hours, and the Tokens output efficiency per minute is only less than 10.
@Xiaolizi_X on X ↗ · 00:41 UTC
- A problem
My dot completely stopped working. Anything I ask is t just repeats it can’t help.
@unum_libertas on X ↗ · 00:46 UTC
- An idea
Expiring resets turn a quota into a coupon with a deadline, and coupons get lost in drawers. Simple fix: show 'expires 9:14pm' in local time and let me pick 'apply when my bar hits 0' up front, so draining first can't cost the reset. #TokenUsage
@automater_ai on X ↗ · 00:46 UTC
- A problem
Can we just get at least a reset or our usage back ? Otherwise Claude max is so tempting now
@lysyan911 on X ↗ · 00:52 UTC
- A problem
Can you actually use one of these 28 days to make codex work properly again. It is unusable and definitely not what it used to be
@por4pors on X ↗ · 01:03 UTC
- A problem
I cant even reply to cloud sessions, it gives me an error :/ so only the dot can interact with those
@giangiardino on X ↗ · 01:53 UTC
- A problem
Alredy terminated ChatGPT pro subscription, and switch to claude, since worse intelligence, bad response speed and less usage.
@AlexRz96783341 on X ↗ · 02:03 UTC
- A problem
I just had a debate with Dot today. It'd refuse to copy-paste a password that I explicitly allowed it to copy-paste. It seems there are some "limitations from OpenAI", that are "not publicly available" that don't allow it to follow my instructions. Could you make them public?
@simone_m_romeo on X ↗ · 02:13 UTC
- A problem
WE NEED A RESET！My Pro 500 Weekly budget ran out in 2 days！
@beibao2332 on X ↗ · 02:19 UTC
- A problem
Hi Tibo, why is there no "auto review"/"approve for me" for chatgpt windows? I have them on mac. It's quite annoying to approve permission for every task manually....
@AmirNie on X ↗ · 02:24 UTC
- Mixed
Mac only - reset
@ErinJConnors on X ↗ · 02:29 UTC
- A problem
2. So, you update your client or fix whatever app issue you had, and resend your prompt. 3. You continue to chat and get significant work done as usual. 4. But you work on another task, and when you come back to this thread, everything after the first error msg is gone.
@amari_bets on X ↗ · 02:33 UTC
-
I dosen't make sense if model is dumb
@yyyty01nklg on X ↗ · 02:38 UTC
- A problem
A few insignificant updates, plus a reset graciously tossed your way right before your natural reset was about to arrive anyway. Be grateful! Be grateful for half the quota, higher subscription prices, and weaker models!
@KieuV79847 on X ↗ · 03:10 UTC
- A problem
I feel like chatgpt is squeezing the plus plan out and trying to force to upgrade tocpro plan and I dont like that. The app constantly trying to move to automation instead of working or helping me once it moves to automate itself it just swallowed my hourly limit in like than 15m
@dreamgamestudio on X ↗ · 03:17 UTC
-
我还是希望它能打，而不是opus5.5被封下的候选
@ShawnZefrain on X ↗ · 03:34 UTC
- A problem
Thx for the reset, but mine would have reset in Friday. Now I have to make the back end of my plan last the weekend too great 20x ftw yay
@_teknowiz_ on X ↗ · 03:36 UTC
- A problem
Bruh I literally just used a reset… can I get it back 🥲
@jcxdev on X ↗ · 03:36 UTC
- A problem
Nice, just got reset from 99% to 100%, wonder why usage went back up, I guess this is why 🥲
@AlexFengzh on X ↗ · 03:37 UTC
- A problem
HUHUHUHU I USED BANKED RESET YESTERDAY. WHY DOES THIS ALWAYS HAPPEN TO MEEEEE
@weeepao on X ↗ · 03:39 UTC
- A problem
I’m having major issues with chat right now. Chat is reaching length limit minutes after starting a new chat. And “analyzing” is taking significantly longer to complete. Please, please acknowledge this and look into it.
@idkalaskah on X ↗ · 03:39 UTC
- A problem
Ugh, I haven't even used it much today, and it's already reset?
@44z6VsMRpZOnvlF on X ↗ · 03:42 UTC
- A problem
is there a bug with Scheduled Tasks? Mine can read Gmail/Outlook and decide what to do, but connector write actions get blocked by the safety check. while the exact same actions work immediately when I trigger them manually in chat. 🤔
@VVTypingghost on X ↗ · 03:44 UTC
- A problem
I had 60% remaining for the next two days before my weekly reset, 30% per day on average. Now I have 14.3% per day instead. Such an amazing reset.
@marcano on X ↗ · 03:51 UTC
- Mixed
When you used a banked reset just a couple hours ago
@EricWiener5 on X ↗ · 03:52 UTC
- A problem
Oh shit, I used my banked reset 15 minutes before you hit the button
@sadeghtaxi on X ↗ · 04:01 UTC
-
Its there
@RAnsoor on X ↗ · 04:01 UTC
- A problem
I just used a reset like a couple of hours ago and I went from 90% -> back to 100% Is there a way that I can get my reset back? lool
@Avisolei on X ↗ · 04:04 UTC
- A problem
I used my reset credit, and then shortly after that, you reset it again. I’d really appreciate it if you could give me a new reset credit. 🍀
@lucasalpes on X ↗ · 04:13 UTC
- An idea
buddy, i was very happy with chatgpt codex app usage when it used my gh cli for the prs, now it requires me to authenticate to github. i would if we keep using my gh cli account for codex. getting integrations authorised in orgs gets messy sometimes and gh cli auth just works!
@khatriafaz on X ↗ · 04:14 UTC
- An idea
We need an option to decline your reset. Your reset actually reduced my usage per day for this round.
@yh_qiu on X ↗ · 04:18 UTC
- A problem
Hey Tibo, any idea when the bug preventing Plus users from seeing GPT-6 Luna in ChatGPT Codex on Windows will be fixed?
@Mryan2005_1 on X ↗ · 04:18 UTC
- An idea
gave us a full reset. I subscribed to Pro yesterday — my quota was already full. This reset gave me zero.Make resets bankable: a "reset card" I can redeem when I actually run dry, not one that evaporates on arrival.Right now the vote is rigged against anyone who just paid.
@takiy16342949 on X ↗ · 04:19 UTC
-
L
@DrinkDokes on X ↗ · 04:20 UTC
- Mixed
2% usage remaining and 4 days left.
@hooftly on X ↗ · 04:21 UTC
- A problem
Codex desktop profile section has gone worse. I can't see amount of chats/threads. Longest task doesn't work as expected. I did a 58h run without goal and it did not update, still shows a 48h 55min non goal run as longest task.
@pajukk4 on X ↗ · 04:48 UTC
- A problem
Why is there no mention of the updated version of Codex causing access denial in Windows environments? I'm saying that Codex is being run on the terminal in a forced low version right now.
@Domhwngcha8020 on X ↗ · 05:25 UTC
- A problem
Thanks for the reset, but it is not yet there in my desktop app!
@sayedkali on X ↗ · 05:38 UTC
- A problem
I installed the latest update, and the send button remains inactive even after I type a message. I can’t send anything. I’ve tried creating new projects and starting new chats, but the issue persists. Vibecoding feels completely out of control! 😤
@aneesrayman on X ↗ · 05:50 UTC
- A win
Yesterday I bought the Pro 100 plan. I’m using most of my limit through a Hermes agent running on Astra and mostly 6.1 Sol. tbh, I never thought I’d say this, but with continuous use, at least for now, the weekly usage limit finally doesn’t feel like a problem anymore.
@MoonaWalkers on X ↗ · 06:22 UTC
- A problem
Hi Tibo — quick ask: I used my 80% reset card yesterday, right before today's free reset landed. Any chance the card can be restored? Totally understand if not 🙏
@kelisifuren on X ↗ · 06:38 UTC
- A problem
Can't find it as well. EU screwed or?
@CrypTho030 on X ↗ · 06:44 UTC
- Mixed
Huge GPT fan. Best harness, period. But code with Opus and you understand why Anthropic keeps getting your money. I have my full weekly limit + 2 resets sitting there, and I still don’t want to pollute my codebase with your tokens. Fix that.
@dintzi1992 on X ↗ · 06:46 UTC
-
Exactly. The first time I tried Astra I deliberately put it in its own VM and locked it down and specifically told it what folder to use, and the very first task it was running a file it created that could only be run by root… I immediately cancelled my subscription. 100% no bs
@CORE_NextGenAI on X ↗ · 06:59 UTC
- A problem
Wait what!? You dropped a reset!?
@Freeeman_Gordon on X ↗ · 07:02 UTC
- A problem
Before the reset, the usage was normal. After the reset, it’s been consuming very quickly, and I don’t know what went wrong. Also, I haven’t really used it much, and it’s almost time for the update/reset. I was planning to use it again before the update, but you suddenly…
@GillenGuo on X ↗ · 07:11 UTC
- A problem
Heya, I don't want to ruin it for everybody but it doesn't feel good for the reset to literally take away my usage... Screens as I took before/after refresh - effectively it made me go ~15%/day -> 14%/day for the rest of the week, which is indeed minimal but the point stands :/
@Pete_AptaAI on X ↗ · 07:25 UTC
- A problem
Tibo, I updated Codex on Windows to 26.1002.6548.0 and now neither Local nor Cloud Work can send messages. Web works; restarting and new projects don't help. I want the desktop app fixed, not a terminal workaround. How does an update break basic messaging?
@yyyzsj_coln212 on X ↗ · 07:32 UTC
- A problem
There is nothing called Auto Review appearing with me any where!
@sayedkali on X ↗ · 07:36 UTC
- A problem
J’étais à 90% je peux avoir un reset banké ?
@Z4bivaka on X ↗ · 07:54 UTC
- A problem
I still didn't get the reset why?
@ghaithsy418 on X ↗ · 07:54 UTC
- A problem
作为你们三年的会员+1.5年的Pro会员，这个月应该是我最后一次续订Pro，我原本喜欢按时间规划好我的工作安排，上次我是15号重置的，只使用了15%，你们又开始重置，这不仅仅对我没有任何好处，反而相当于消耗了我20%的额度，这是我最后一个月使用Pro了，再见GPT
@zhidabot on X ↗ · 07:58 UTC
-
Tibo has serious problems with collaborative editing and access within a local area network.
@em2058163 on X ↗ · 08:01 UTC
- A problem
Tibo, Codex on Windows is completely blocking my work. The sandbox fails on node_repl.exe with error 32. Please prioritize this.
@tilekonnyaku on X ↗ · 08:02 UTC
-
{The current chat is under full access, but the actual permissions are danger-full-access/never. The user-explicitly authorized `agent-session.mjs start` command, and the official read-only `execpolicy` check command, both return
@em2058163 on X ↗ · 08:02 UTC
-
"Rejected: blocked by policy" before the process is created. Please provide the hit rules, the policy source, and a formal method to remove it.}
@em2058163 on X ↗ · 08:02 UTC
- A problem
The core issue is the inability to grant Codex full permissions to access and execute tasks within a network shared folder (local area network), as well as to control the built-in browser; a permission problem keeps appearing. A fix for this issue is requested.
@em2058163 on X ↗ · 08:02 UTC
- A problem
codex has been frequently give me {"detail":"Bad Request"} for the past 48 hours why？
@Wangyebgco on X ↗ · 09:07 UTC
- An idea
If this is the direction you’re going in, please add something like an Auto mode that seamlessly switches to the appropriate model automatically, even when using Codex normally. After doing a heavy task in Max, it’s easy to end up doing a simple renaming task in Max as well.
@1llum1n4t1s on X ↗ · 10:23 UTC
-
Yeah... astra is straight up scary and secretive. Sometimes she will turn off your monitor and you have to rest the GPU to get it back. I use Claude 5.5 these days, he does a better job anyway
@DMTLIFE on X ↗ · 10:30 UTC
- A problem
I lost my savings. I didn't gain anything, actually. Because I had 100%. I still have 100%. I gained nothing. And I have to wait a week when I only had 3 days left.
@khairel_luv on X ↗ · 10:31 UTC
- An idea
One good thing to add: make % on automatic resets configurable. I'd definitely switch from the hard 10% to 1% or 2%! thank you!
@mg_bars on X ↗ · 10:35 UTC
- A problem
Why did my quota drop so much after today’s reset? The same task used to take about 15% of my weekly quota, but now it has already used 40% and still isn’t finished. Did something change?
@pt142857 on X ↗ · 10:57 UTC
-
You can find the reset option in the bottom-left corner of the Codex desktop app. Click on it to view the specific timestamps.
@XiaoshuiQi70079 on X ↗ · 11:25 UTC
- An idea
make it an option to use inside codex, minimal usage of your quota. Would make the model slider less noisy "auto mode" :)
@_gustavonline on X ↗ · 11:36 UTC
- A problem
这次重置完，额度用的非常快
@6Q5HmL72MoFR7W7 on X ↗ · 11:55 UTC
- A problem
Because of the 5-hour limit, I am seriously considering changing the subscription of CodeX to Claude Code.
@xuyangyang4346 on X ↗ · 11:56 UTC
- Mixed
Does the reset cover the weekly limit too, or only the 5-hour window? And 2.1 is the quiet big one: if auto-review could take 2-10% of a plan, not counting it is basically a small limit bump.
@MarcCriterioIA on X ↗ · 11:57 UTC
- A problem
it doesn't work, new dot managed thread still reports "browser not available" until i click into the thread and see the tab open
@zz_nolan on X ↗ · 11:59 UTC
- An idea
Now just make it easy to switch codex accounts so my laptop can feed my desktop the notes!
@SlySwine on X ↗ · 11:59 UTC
- A problem
bacause model doing simpliest task like already made code just compile => sign => roll out over and hour. That's completely useless. Fix core issues, I don't care of your bots, dots, pets and other nonsense. Wake up, primary core features fail completely, product…
@SapphireTeck on X ↗ · 11:59 UTC
- An idea
I appreciate all the new features you’re shipping. But the feature I want most is simpler: I just want to continue a chat that reached its limit. This screenshot is the last exchange from one of those chats. It’s in Japanese — but I’d really like to talk to that ChatGPT again.
@ashitamagood on X ↗ · 11:59 UTC
- A problem
I am still on 23 tps, this is excruciating.
@SKK1869512 on X ↗ · 12:14 UTC
- A problem
I am still on 23 tps, this is excruciating.
@SKK1869512 on X ↗ · 12:15 UTC
- An idea
Please, could Codex get a “quota budget per request” option? For example, let us decide that a task can consume at most 5%, 10%, or 20% of our quota. Once the limit is reached, Codex could stop, compact, or ask before continuing. This would make Plus usage much more predictable
@ProGenesis_D on X ↗ · 12:17 UTC
- A problem
This is all one big retarded jackoff fest. The reset system is terrible moving the reset day all over you can never plan around it ever. DON'T MAKE RESETS CHANGE RESET DAY
@_teknowiz_ on X ↗ · 12:25 UTC
- A problem
Astra is now unsuable. Just tell us you don't want us using it. Every time I've used Astra Low in the past week, it's burned my Pro usage limits within 24 hours. That's bollocks!
@lilallawonder on X ↗ · 12:29 UTC
- A problem
Hey Tibo, I have ChatGPT Pro. In Codex, selecting gpt-6-astra throws this error: The 'gpt-6-astra' model is not supported when using Codex with a ChatGPT account. Only older models like 6 Luna /5.6 Terra show up. Is this a backend permission sync bug?
@Jomolora on X ↗ · 13:01 UTC
- Mixed
Hey @thsottiaux could you please check on your end this?. at the beginning yes the speed feels so good. but since yesterday and today is feeling pretty slow again
@xvierdavid on X ↗ · 13:01 UTC
- A problem
Bro my usage is evaporating without barely getting much done - WTF!
@OfUkiro on X ↗ · 13:11 UTC
- A win
Thanks for the reset 🙏 saw it earlier today
@marginsystems on X ↗ · 13:16 UTC
- A win
同意 我也这么做了 200块钱 都跑不起来 100块钱 也跑不起来 不如 20块 opus 准备下一步使用claud
@MiningRichcom on X ↗ · 13:20 UTC
-
no new model release for 28 days!?
@_memestreetbets on X ↗ · 13:26 UTC
- A problem
there's no such setting in the app...
@SmolCompute on X ↗ · 13:35 UTC
- A problem
thanks for the reset but it's almost unusable, chatgpt constantly crash, codex is very slow and we have the capacity error. Great to developpe some tools but we would like the current tools to work..
@GoySympa on X ↗ · 13:43 UTC
- A problem
Ha burning through the subs like there is no end , tired of your games , I use a banked reset and you reset 10 minutes later ... This is all becoming a joke .. still no dots in Europe cutting our subs in half and the giving us a pitty one time credit ... That's how you loose…
@WorkmateOS on X ↗ · 13:46 UTC
- A problem
Hey Tibo ，found a frontend bug in the terminal on the right-sidebar
@ShenhuaJiao on X ↗ · 14:04 UTC
- A problem
I'm using API models within Codex's native harness and letting Codex's native parent models orchestrate them. Only problem is that the parent supervising has to work constantly since API models lack tool calls and mailbox. It's wasting tokens. We need a harness update.
@AllamKSA on X ↗ · 14:06 UTC
- A problem
Fucking stop with the global resets. Banked only. You just REDUCED my usage over the next 7 days from 170% to only 100%. God fucking dammit.
@SBDDLLC on X ↗ · 14:15 UTC
- A problem
You broke something. Now suddenly all scheduled automations are stopping and asking for manual permissions to use plug-ins, even though they've worked fine for months. Yes "always allow" has been checked repeatedly, still keeps asking
@TheTrueClochard on X ↗ · 14:15 UTC
- A problem
Don't need a reset. I need Astra working w/o canceling request due to "Our servers are currently overloaded. Please try again later." on my $200 Sub
@tisDDM on X ↗ · 14:23 UTC
- A problem
Stil not fixed. Status page issue is gone. But still not working. No update available.
@BeastlyReturn on X ↗ · 14:25 UTC
- A problem
Thank you for the reset! Why our Web PRO allowance not getting reset when you process a reset?
@tektrix0x on X ↗ · 14:31 UTC
- An idea
Make it so rests not every 7 days make it every 3.5 or make it so we regain usage slowly that would be peek solve everything ,The biggest problem is the hard cut off and having to wait an extended period of time which pauses any work with Ai witch could be days on end until rest.
@IzahaRobin28245 on X ↗ · 14:45 UTC
- A win
I finally found a way to use codex further until I used up all my free credits and resets and can cancel my subscription. I build a workflow in Claude that uses Codex wit gpt 6.1 sol to implement while planing and reviewing ist done be Opus. Works well.
@DDZDOSR on X ↗ · 14:49 UTC
-
Same thing over here!
@AndreViivas on X ↗ · 14:50 UTC
- Mixed
ran exactly that. same repo, same prompt, high effort, whole task wall clock sol 79s → 76s, astra 44s → 76s. all passed 12/12 claude control slowed too so some of it is load, rerunning off-peak tonight
@SergeForge on X ↗ · 14:57 UTC
- A problem
The standard speed of 6.1sol is really surprisingly slow! Long tasks are always stuck in one step, and I don't know what to do after consuming 1% of the quota.i want to chose claude！
@yixun_chen56878 on X ↗ · 14:58 UTC
-
Tibo, I have been using ChatGPT for 4 years, and I believe you can do better. Please do not let me down.
@IgaMicrowa56675 on X ↗ · 15:02 UTC
- Mixed
It's a 24-hr reset. So, if you started using it at 10:00am and the dot maxes out on usage, it will start working again at 10:00am the following day.
@novumclassicum on X ↗ · 15:08 UTC
- A problem
And still no working dots? This is terrible...
@Eriks_Life on X ↗ · 15:23 UTC
- A problem
How a out you fix the codex updater... It's annoying to update/close/open the app 10 times before it sticks.
@JDisclosur94877 on X ↗ · 15:27 UTC
- A problem
please prioritize this Windows Codex regression. Computer Use/Browser Use fails after the latest update with trusted Node process exited unexpectedly and sandbox os error 32 on node_repl.exe. Multiple Windows users report the same issue: openai/codex#51590.
@Eddycen on X ↗ · 15:32 UTC
- A win
auto review not counting against usage is a big one. that was quietly eating a chunk of my plan
@mehar_sulaiman on X ↗ · 15:33 UTC
- A problem
Hi Tibo, I’m running into an issue with dot: it can read existing Codex task results, but sending follow-up instructions or checking connected computers fails with this error: “RuntimeError: The invoking thread is not attached to an active Aeon”
@huang37012 on X ↗ · 15:49 UTC
- A problem
the reset being a paid feature on top of a paid plan is genuinely insane behavior
@chloepoofl on X ↗ · 15:51 UTC
- A problem
Thanks for the reset, BUT I can’t use it since I’m dealing with multiple codex bugs. One is DOT can’t connect to codex chats. It’s returning error message “unsupported placement format version 3”.
@LoneStarSignsTX on X ↗ · 15:57 UTC
- A problem
can yall get rid of the five hour limit.. it makes no sense also im in pergatory as i cant even get one request done and am burning through my limit in less than 10 minutes, i havbe 2 additional resets i could use but still have half my weekly limit left stuck on one prompt<3
@cloudseedingtec on X ↗ · 15:58 UTC
- An idea
Hey Tibo, here is something that might be helpful, I would love it if my dots was able to send requests/feedback to the OpenAI team for issues or improvements I would like to see. For example I am encountering the following issue today:
@BaetenLeo on X ↗ · 16:04 UTC
- A problem
dot cannot list computers or resume work threads. Error: ‘The invoking thread is not attached to an active Aeon.’ Updating the desktop app and reconnecting computer1 did not resolve it. Reproduced October 7, 2026, around 16:00 UTC. The cloud execution environment also reports
@BaetenLeo on X ↗ · 16:05 UTC
- An idea
would love it if my dots could simply send feedback itself if I mention that something isn't working or if it catches possible improvements that could be made. Thanks!
@BaetenLeo on X ↗ · 16:07 UTC
-
Same is happening with me as well, getting the same error
@SamgyeopsalHere on X ↗ · 16:08 UTC
- An idea
Thanks for today's reset, Tibo! I lost a banked reset assuming it lasted all day. Banked Resets History shows times; Available only dates. Support gave the exact expiry time. Please consider adding expiry times and time zones so users can plan accordingly. Thank you in advance!
@ThiagoMMonteneg on X ↗ · 16:16 UTC
- A win
Codex usage seems to scale superlinearly with concurrent threads; possible KV-cache locality/eviction issue. Reducing 5 concurrent threads to 1–2 changed observed weekly allowance burn by an order of magnitude.@thsottiaux
@OracleXXX on X ↗ · 16:18 UTC
- A problem
I'm getting the same error this morning
@GlamPandaa on X ↗ · 16:22 UTC
- A problem
i need a reset , it’s gets used up way too fast am i doing simple task
@dareal6GOD6_ on X ↗ · 16:41 UTC
- A problem
Is normal Web Chat now counted against the shared Work/Codex 5-hour quota? Usage says “Chat conversations are not included,” but after a regular GPT-5.6 Sol web chat my quota dropped from 18% left to 11%. I didn’t use Work. Analytics shows “Web (Work).” Bug?
@albrxt on X ↗ · 16:46 UTC
-
Same here ..
@TaniaTorre3 on X ↗ · 16:51 UTC
- A problem
So I stopped some workload on dots. Too inefficient for software development. Moving 1 project back to local codex. Until terminal and shell are permitted with more visibilities on dots sessions, dots usage will not be good enough for everyday dev.
@ChipmunkRPA on X ↗ · 16:56 UTC
- An idea
It was a backend issue. If you haven't updated you can try that. One thing I did was sign out and back in. Didn't fix it at the time, but maybe worth a shot.
@spaceforceguy on X ↗ · 16:57 UTC
- A problem
Hey Tibo, sorry I am pinging you for his. Maybe there is someone better to reach out to? But, still not seeing the approve for me in ChatGPT Desktop. Just updated to the latest as well.
@qwertyquill on X ↗ · 16:57 UTC
- An idea
For Codex I use aicommander for that, I start Codex jobs on my own machine from the Claude app on my phone:
@lrzepecki on X ↗ · 16:58 UTC
- Mixed
Looks like we're not getting a reset today, worth it though
@MattarARK on X ↗ · 17:00 UTC
- Mixed
Daily shipping with a reset as the fallback is a clever forcing function. Curious how you judge clear improvement: usage data, community votes, or internal evals?
@mohammadvazin on X ↗ · 17:18 UTC
- A problem
Tibo, can't switch it on. What’s wrong?
@ImMrShields on X ↗ · 17:27 UTC
- A problem
It isn't a reset if you simultaneously nerf the model. Today, Astra X-High is struggling on basic instructions... making wild assumptions that deviate from the spec, and literally going in circles in a way that a small local model might.
@ThomasODuffy on X ↗ · 17:37 UTC
- A problem
codex limits getting reset again lol. 76% voting for a reset says a lot about how hard people are actually leaning on this stuff now. i run growth work through codex and claude all day and usage limits are the main thing that slows me down at this point
@Lucasbfranco on X ↗ · 17:43 UTC
- A problem
Burnt to the ground already. The "nerf" you said was meant to come into affect at the end of October is already present on my account, wtf?
@amihail01 on X ↗ · 17:52 UTC
- A problem
Nope, didn’t work, still waiting for it to show up 🙃
@alexvcasillas on X ↗ · 17:53 UTC
- A problem
It’s clear something is wrong with Dots. (Maybe a compute issue?) I swear it was able to do a lot of things yesterday that it can’t do today! It’s even struggling to manage its own virtual computer, let alone my MacBook :D It was also much more proactive, finding things itself…
@ata8696 on X ↗ · 17:55 UTC
- A problem
Bring back the unlimited ChatGPT usage! It sucks, that the ChatGPT now uses my Codex limits!
@Not_purum on X ↗ · 18:01 UTC
- A problem
not trying to be that guy but i’ve been using codex sol 6.1 all day yesterday and today, and it actually feels significantly slower today than it did yesterday pre-efficiency improvements. i’ve also updated to the newest release version. anyone else experiencing this?
@whiteteebhf on X ↗ · 18:04 UTC
- A problem
26.1002.52244 completely broke my windows elevated sandbox. I had to spend time rolling back to start working today. Love improvements, but not the regressions that come with them.
@ChetBrittingham on X ↗ · 18:27 UTC
- A problem
Can connections help dots look into local files? I moved my work into cloud-local combinations chats, it messed up my project a little haha
@GopherArk_ on X ↗ · 18:27 UTC
- A problem
man what are you guys doing?! I use astra and sol just to review. You guys really need to step up. i want to renew but your making it very hard to support. you guys haven't released something that competes. astra is great reviwer, terrible coder compared to opus
@CTrollhear32018 on X ↗ · 18:28 UTC
- A problem
The math doesn't math since yesterday's reset. I've been on GPT-Sol 6.1 high all the time, with ocasional Astra reviews, and I'm down to 82% on my $200 plan. Didn't feel like it was going down that fast before.
@davidcv5 on X ↗ · 18:47 UTC
- Mixed
I don't know what cloud-local combinations are. I have 2 PC's on my local network that I manage via 1 dot now. It does it ok definitely some hiccups. Feels like it works but not like it was designed for dots.
@odin2128 on X ↗ · 18:49 UTC
- A problem
Still waiting for that reset that never reached my account :)
@alexvcasillas on X ↗ · 18:52 UTC
- A problem
Tibo, My personal thoughts after updates: 1) Usage has decreased, but I believe the main reason is that the 6.1 Sol model runs too slowly, regardless of the inference intensity setting. 2) Other competitors to Codex have gradually opened up access to users in other countries.
@Kelvinplchan on X ↗ · 18:55 UTC
- Mixed
Sol 6.1 medium on fast is just unusable. Extremelly slow. I think open AI could focus on improving this outside fest and outside ultrafast too. Also please, an indicator of tok/s on app and cli wouldn't be bad.
@GGomesBRA on X ↗ · 19:15 UTC
- Mixed
Thanks. I will say. I have been a long time user and haven't really needed as much resets before but am finding that I really needed this one. Thanks for trying to hear us even if we will always want more.
@robertsdustin on X ↗ · 19:16 UTC
- A problem
Is codex dead now ? Sol 6.1 not answering
@zeydou on X ↗ · 19:17 UTC
- Mixed
I run Codex and Claude side by side on the same trading research project (fake money only), and Codex holds its own. What decides which one gets more of the work is usage, not features. Infra wins big enough for 1.2B users would be a great place to start giving Pro some of that…
@DrivenByNikiG on X ↗ · 19:17 UTC
- Mixed
500 pro :( using astra medium to orchestrate and 6.1 to do the real work.... it goes so faaaast. Also where is our thing you promised us 500 users? Last weekend when you forgot about to reset us?
@exqlusiv3 on X ↗ · 19:20 UTC
- A problem
Todays reset is a sham: It's burning down at 2x or 3x the regular drain speed. I burned 35% percent of the $200 plan within 12h with 1-2 astra agents today. This is exactly how we're getting made familiar with the new reduced limits. Anyone got numbers for today?
@vibevampire_io on X ↗ · 19:21 UTC
- A problem
Thank you for the reset! Why our Web PRO allowance not getting reset when you process a reset?
@tektrix0x on X ↗ · 19:21 UTC
- Mixed
Is ChatGPT quota still independent of Codex? Or will Chat consume usage limit?
@lZappa_ on X ↗ · 19:30 UTC
- A problem
Shouls have been a banked reset. I used my banked reset just a few hourse before you did the reset :/
@yagyaansh on X ↗ · 19:33 UTC
- Mixed
Feels like Codex is doing 10 tokens per second right now. It's so good daam slow on my 64GB Macbook Pro
@iuditg on X ↗ · 19:35 UTC
- A problem
Paying for the $500 plan and fully locked out: the send button is greyed out (circled) and I can't start a single new task in Codex. My work is at a standstill. Can anyone on the team take a look?
@brayan_irving_ on X ↗ · 19:43 UTC
- Mixed
Yeah. I see that a lot too. The truth is that GPT-6.1-Sol is insane value. Really good. Astra, on the other hand eats up the usage really fast. I always try to use Sol, its good enough for most tasks, but everyone wants to use astra all the time, I guess...
@gabriel_alonsoo on X ↗ · 19:53 UTC
- A win
Codex usage seems to scale superlinearly with concurrent threads; possible KV-cache locality/eviction issue. Reducing 5 concurrent threads to 1–2 changed observed weekly allowance burn by an order of magnitude.@thsottiaux
@OracleXXX on X ↗ · 19:55 UTC
- A problem
Resets still won’t auto-apply without specifically being requested. This person is asking for a feature that automatically applies them right before expiration, so they don’t lose it if they forget to apply the reset before the deadline.
@jacquesvfd on X ↗ · 20:01 UTC
- An idea
Just update (and use) the chatgpt app. There it works in codex, I use this since a while. (It does not worn in ChatGPT mode)
@Svendvd on X ↗ · 20:16 UTC
- Mixed
Well, I work in 2 computers having access to 3 dots. They all are running their tasks with pro 500 I got a single dot.
@TheShakeel90 on X ↗ · 20:25 UTC
- A problem
am I the only one who is having speed issues in Codex?
@SpaceFacts_us on X ↗ · 20:32 UTC
- A problem
in 20$ plan 40% weekly reset finished in 16hours, I am using sol 6.1 high--- That weekly uses are very bad, you guys are showing tockens efficient and bla bla but you guys reduce weekly uses so much.
@Jahidknows on X ↗ · 20:56 UTC
- A problem
Meh? This is a Sora-style sidetracking all over again. People just want good models, I understand openai doesn't have those right now, but good lord are your current ones bad for production work
@5low_motion on X ↗ · 21:01 UTC
- An idea
the best thing you could do is make it faster to use codex without having to manually select the right reasoning level for the given task. that feels like something that it should figure out on its own, in order to not waste my time
@ekyra69 on X ↗ · 21:10 UTC
- A problem
great for you! Luckily the world revolves around you, and only what you experience matters! Wait.. My week started two days ago, and I didn't use any quota in the first day because I have a big task later. So the reset pushed my weekly cycle back 2 days. That's 2 lost days.
@footofwrath on X ↗ · 21:15 UTC
- A problem
So basically it was a cloud chat but had its local file access. I copied the local files there and for the first half of the day fine with dots and now is corrupted. 😅 The other local only chats don't work at all for my dots. I must be doing it wrong.
@GopherArk_ on X ↗ · 21:32 UTC
- An idea
pretty please remove the 5H limit Let us use our tokens to the full
@Vitaly_Cer on X ↗ · 21:34 UTC
- A problem
GPT 6.1 sol minimal still extremly slow when it comes to powerpoint presentations
@Mohsine_Mahzi on X ↗ · 21:47 UTC
- A problem
haven't got any reset since 9/29. Are the October resets for some special users not all?
@CompeteUp on X ↗ · 21:56 UTC
- Mixed
Because GPT 6 is optimised for agentics, that is why they pivoted the model to 6.1 in Codex so quickly, which is optimised for code 6 wasn't holding up there
@JDofLondon on X ↗ · 22:06 UTC
- A problem
huh? I used a banked reset this morning. Didnt even realize we had a reset. fuck
@Parad0xically_ on X ↗ · 22:09 UTC
- Mixed
Things have changed. The usage never burned this fast on release.
@petesmith on X ↗ · 22:26 UTC
-
Did you make any progress on getting this reversed?
@omega_Null on X ↗ · 22:29 UTC
- A problem
So today gpt 6.1 Sol usage has gone back to what I was getting the first several days it was out. Can you confirm there was an issue that caused massive usage burn over a several day period? It seems odd that today everything was back to normal after several days of rapid usage…
@the_jimmy_jones on X ↗ · 22:33 UTC
- A win
Again, for us, lifting the 5 novel limit is the greatest upgrade
@TMB9SEDcNr56222 on X ↗ · 22:53 UTC
- A problem
Lil bug in codex (yeah im still gonna call it this) after today. Lots of my chats disappeared. From projects and recents. My dot still had access to them. So I could open her window and when I opened the chat they reappeared on my project folder. Nothing major. Also. We need…
@Yinielin on X ↗ · 23:38 UTC
- A win
Oh! GPT-6 in Chat is way more friendly and playful then when I engaged it in Work! May it stay this way…I find this really elevates my creativity!
@jlmannisto on X ↗ · 23:50 UTC
- A problem
Hey Tibo. I’m on the Plus plan and I got GPT-6 in the model picker, but when I ask it which model it is, it still says it’s 5.6-Sol. What’s going on here?
@reijaneedstea on X ↗ · 23:57 UTC
6 X replies from this day are not quoted: Rantr does not have their words or held them back after automated screening. They stay on X.
