My Codex weekly reset appears not to have applied, or usage immediately returned to 100% with no active work.”

Potential duplicates detected. Please review them and close your issue if it is a duplicate.

• #31402
• #31415
• #30206
• #30128

Powered by Codex Action

as we use a new tool ,, next we use it more and more

I seriously recomend you using 5.5 medium - as its unnoticeable in many cases
and faster, and you get higher usage
I personally I think its highly debated --- how the codex app - should actually be treating your codebase or whatever --- as things stands now - all of the llms takes some how 500 lines there and there... it takes functions ( Bilbo Baggins: "We hobbits are plain, quiet folk. Adventures make one late for dinner" ) in a sense i mean - you usally have 5 files in your mind generally understand what they mean and what for (and if you are working for a company you maybe have a plan-goals --- or ideas on how to prevent link-rot [which amazingly I think started to happen less after ai was a thing -- but its happening , as a lot of foul practices - and i dont know how I feel about it --- in a world where there is dozen of standards - I dont understand why landing pages starts to break on such large companies - like how they keep internal recipts or whatever - and its seems like the world is perfectally ticking when a big chunk isnt fucntioning great --- you know

[random old man bubble by me ]
there is some trends you hope would continue -- like good dentures being made by some kind of physical-ai , which is a complex surgery -
--- but there is plans, a roadmap or whatever --- a good doctor who does all-on-4 cant do 150 patients a year -- and not 900 patients over 10 years... when clearly there is a need for a lot more people being on "all on 4 " dentistry is actually an indurstry with a lot of malpractic --- and you know a lot of people would want to functioally bite again ] ) -- I recently got my 86 year old dad on {dental dentures}all-on-4 and it was costly (but I can afford it - but not for more than 40 people -- if we calculate the econimics of things) -- and if we had a sudden brust of free-goverment-funding , the only thing that would happen is for the surgery would becoming more expensive because - its very rare to have a good doctor .... (and seriously what is this. remove-gum and bones ---> consideration of quality of bone [soemthing that doctors fail often -- and ai can automated atleasat the part of CONSIDERATION && EXECUTION OF THE SURGERY] - I guess in coopate software developement , a roadmap//meeting (i am from the mining indurstry so I am just a consumer!) and honestly as things stands down --- we only need a very good CT scan --- and getting good data and how to drill bone quiality or whatever --- because the 100+ patients could become 700 patients a year --- and a lot less malpractics which I call it 2-to-4 ratio of bad failing surgeries after 4 years ---- and then you have to consider how patients lie on what they do post surgery . . . but you know its practical I personally thing ai-robot-surgerys to be 3x times cheaper ----and then the goverment/company[maybe to get accepted by law for the practice of such surgery] would be limited to cover 2x times the failing-rate so twice the failing-rate in lab settings ,,, and needs to be 4x safer ---- and you know because its about drilling and bone quiality (and then CAD design) I actually think its very different than doing heart surgery or replacing lungs --- which is more of a soft tissue thing ---- but as per denistry considering how bad the state of things , its very important to have some kind of solution --- because a lot of people going to suffer from bad dentists and serious malpractice --- and its more about losing a tooth, of more about lossing a lot of gum and bones - and you can be toothless you just need to be able to chew on something , orally function... not for some random nobody to drill into your face and destory bone you cant use again --- and some do it ,, and they have insane amount of pain --- when a good doctor could have done it the proper way [even if both implants will survive for many years] --- which is funny ) and what funny is that ai could like get to 95% in the currect place to place an impant -- without paint ,,, while not a surgery to connect a finger and nerves (which takes hours for a human) -- I personally think it can actually done the right way....
as we will apporch the 10 billion people in the few years, I think its a very nessiary things to have some kind of automation --- before seen deaths of mass scales we did not see in the black plauge and middle ages --- (in what cases laws get adujust , and maybe in the future there is an inflation of laws ---- maybe law would become a linguistic issue -- and law should be simplified to linguistic logic --- as automation is an harder issue the world is facing right now --- which you can tell the world is functioning more or less ... you can even be a tourist for few weeks and come back few times a year - most of the people are infact able to eat - if had a little bit of luck -- those in bad luck sadly stay in bad luck --- I personally dont mind low birth rates -- because if we had a baby boom like in the 1950s today, we would been 30billion people today --- and quicker than not --- can we really afford more people in china or singapoor without creating much more suffering, even in south korea -- south korea was 20million people in the 1950s now they are 50 million.. and its not inherintally bad if it drops much lower ,, and you know I had few kids and its one of the highest quiality of life I had --- I dont want 12 billion people all the sudden to suffer -- as per vacines - it used to be taken 8 years to get a safe vaccine --- but now the development and craft and mechinism is pretty much known, I am sure there is a talent to mass create a vacine if incase we get something like ,,, a disease in a 4,000 chickens farm -- and all the chickens die all the sudden --- or like the black plague killing 60% of people -- for the most parts we can stop bacteria -- for the most part things dont get infected like the flu -- but like you get hiv --- (the two are virus -- but the two have very different mechinisms on how they attach to cells and different type of cells ))

sometimes even
grep -rn "myFunction\s*(" .``` which is a very interesting way of doing things (I actually seen nemotron ultra 3 does it which is very good -- and im sure the grep some how changes based on the coding style too )

, (5.5 medium === 5.4high )

second I also recomend you doing "new-chat" from time to time (I am starting to think if you compacted the chat like 10 times - in the same chat/thread -- it has a tendacy to actually try and think of too many things and starts being off topic after a while --- when lets say you focus on a function/or a feature --- but because you spoke to it on 10 different things - the next message would be a bit slower)

<img width="1344" height="702" alt="Image" src="https://github.com/user-attachments/assets/579d0db9-bbae-4e60-998b-e2dd2bb66aba" />
<img width="298" height="313" alt="Image" src="https://github.com/user-attachments/assets/29dadf96-4df2-4108-a31c-f0a2d85a2a40" />
I marked nemotron has an honrary mention because its a very cheap model--- that being said - I dont think human who can built a nice website --- I guess the human could utalize basic chatgpt usage (resize the button, do this and this) , and the human would still burn with less than --- the other way he just used ai f
or basic request - and some help from the chatbot -- yet he didnt spent 10 million tokens --- what I want to ask is , what do we do with problematic tasks , like nodejs/nextjs/nuxtjs webdevelopment --- which consume a lot of tokens for some reason (I personally find it funny to toy with it, and keep telling ai that its wrong -- but its not funny anymore because it now stops - and also explains why its actually currect) - I personally think gpt 5.5 medium is extremely good - it also stops faster for some of the task (what I dont like is - that it tries to code a lot more -- but the code is actually very solid - I actually like to keep the code base smaller) ---- and I think its beautiful I think its long the days where tokens when single letters - but words, topics , etc .... in some ways you can still use it for a simple task - and it will argue it cant do it --- while it can - while you need to tell it use another source or whatever

Update / correction:

I may have partially misread the usage UI. The panel says "Usage remaining", so 5h 97% and Weekly 100% probably meant I still had plenty of quota left, not that usage was exhausted. The VS Code status bar appears to show the opposite style of number, which may have added to my confusion.

I have also found what may be the more likely cause of the stalled Codex turns/processes: Windows Defender repeatedly blocked/removed the Codex Computer Use helper as:

Trojan:Win32/ClickFix.DE!MTB

Affected file:
C:\Users\<user>\AppData\Local\OpenAI\Codex\runtimes\cua_node\1b23c930bdf84ed6\bin\node_modules\@oai\sky\bin\windows\codex-computer-use.exe

My local checks:

• Authenticode status: NotSigned
• SHA-256: F2B2F56FCD1699B0FA32DEC3214A56A1D36B937A2ECF58CC822AB4A904551E03
• File size: 1,691,648 bytes

This appears to match the existing Codex issue here:
https://github.com/openai/codex/issues/31419

So my revised view is that the quota/reset issue may have been a UI interpretation mistake, while the actual stalls may be related to Defender removing the bundled Codex Computer Use helper. I have not allowed the Defender threat and have not added broad exclusions.

Update / correction: I may have partially misread the usage UI. The panel says "Usage remaining", so 5h 97% and Weekly 100% probably meant I still had plenty of quota left, not that usage was exhausted. The VS Code status bar appears to show the opposite style of number, which may have added to my confusion. I have also found what may be the more likely cause of the stalled Codex turns/processes: Windows Defender repeatedly blocked/removed the Codex Computer Use helper as: Trojan:Win32/ClickFix.DE!MTB Affected file: C:\Users<user>\AppData\Local\OpenAI\Codex\runtimes\cua_node\1b23c930bdf84ed6\bin\node_modules@oai\sky\bin\windows\codex-computer-use.exe My local checks: Authenticode status: NotSigned SHA-256: F2B2F56FCD1699B0FA32DEC3214A56A1D36B937A2ECF58CC822AB4A904551E03 * File size: 1,691,648 bytes This appears to match the existing Codex issue here: #31419 So my revised view is that the quota/reset issue may have been a UI interpretation mistake, while the actual stalls may be related to Defender removing the bundled Codex Computer Use helper. I have not allowed the Defender threat and have not added broad exclusions.

do you actively use the computer-use thing? I understand why you might be confused with the