Limit reset consumed while Usage not actually reset

After resetting my limit I just woke up to 0% weekly usage left.

I didn’t do any significant coding then.

After the reset did you get 100%?
And did it do some work then or was it idle? What you define as significant and what others define as such is sometimes very far away.

You can grab the session id and send that to the support.

how do we send to support?

What is your support question? Lower allowance? That is not a bug. That is reduction in service for same price. OpenAI is basically giving API access for free. The 200$ sub in August - if you’d used it all would have given you at least 250,000 $ worth of API requests.

The models are expensive! And you don’t really get much for 20 bucks. That is not a bug, that is basic economics.

there is zero incentive to use codex if after 2 years paying for it two resets less than 40mins of work output. I will rather pay for claude and grok before i ever considering giving openai more money because of this terrible treatment. They get me all excited over this bs i would rather they didnt even give me the feature if this is how they treat me.

I would suggest you use grok and claude as well… But here is another basic economic thing: you get much more for the same price from OpenAI.

What exactly is it that you don’t like how they treat you?
What did they do to you?

how do i get more when a normal chat does better work than codex does? I cannot find any proof after trying to do all i can.

codex runs out of token leaving me a broken code base because of running out of tokens. work is worse with codex.

It takes time to learn how to master codex. And unfortunately also money.
The same as it takes to master driving a ferrari or a yacht.

Could you open the codex app please and check how many tokens you used yesterday?

I can give you some advice where to get tokens for free if you want to. Not the same quality but you can at least learn how to use that stuff:

  1. option: register at NVIDIA NIM … they give you access to their models for free for I think up to 6 months
  2. register at Hetzner Experiments you can get free access to Qwen .. use that in combination with opencode
  3. download kiro - they give you some free allowance
  4. download antigravity - they also give you some free tokens
  5. take part in hackathons at devpost .. somtimes they give you free tokens for the hackathon
    … if you need more just send me a message

i don’t understand how the developers are so clueless too, work is supposed to be the top most of the top, how are they not combining normal chat into what work can do. Use normal chat to do basically tool requests and polling, use sol to do the planning and instructions, then the tokens would be extremely optimized. no more complaints from users.

why is a Ferrari holding the door for a Toyota.

It also costs money to drive a Toyota

you can go further in a toyota and the mechanics are cheaper.

I would rather buy more tokens atm and not waste my time on anything other than running agents.

i can solve this problem for openai but i dont think they want a solution, they want people to pay for more credits.

https://openai.com/careers/ OpenAI is desperately searching for that. Send your application here.

I think you just changed the trajectory of my life, I never once considered this, thank you.

Good luck with that. Take your time with the application.

Definitely not $250,000 worth of tokens. I have the $200 and I think it s probably around $10k/week so abt $40k

30/70 in/out .. with a mix of sol, terra and luna and 1.2B tokens on average per day on my side would be more like 315k

But of course with alot of cached input and 90/10 in/out that would be more like 40k… still worth it and still basically free.