Still burning tokens very fast with today's VS Code extension update

[cy-ooi88]

What version of the IDE extension are you using?

26.311.21342

What subscription do you have?

Business

Which IDE are you using?

VS Code

What platform is your computer?

Microsoft Windows NT 10.0.22631.0 x64

What issue are you seeing?

Am I the only one still seeing my tokens burning very fast following today's extension update? e.g. just by writing 1 or 2 prompts, usage drops by 1%, within 2 hours of working I managed to burn through ~20% of my tokens. Tried using GPT5.3 and 5.4, High reasoning effort. I know High reasoning effort will consume more quickly, but is it considered normal to consume one fifth of the tokens within 2 hours even for business accounts? Doesn't make sense to me, but please clarify

What steps can reproduce the bug?

just by writing reasonably simple prompts. GPT5.3 and 5.4, High reasoning effort.

What is the expected behavior?

Last week token usage was like 25% slower.

Additional information

No response

[github-actions]

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

• Rate Limits dropped 30% within a few hours of use? Teams account - not getting usage resets and still getting fast drops #14349
• Regression around 0.111.0: usage drain spikes with Fast mode default + more aggressive multi-agent spawning #14116
• one prompt consumes 11% of weekly limit #14441
• "Weekly limit % left" dropping suddenly without any usage occurring. #14023
• usage is very high on light usage #14112

Powered by Codex Action

[cy-ooi88]

Not a duplicate. I am aware about the separate reports that are now closed. But with my VS Code extension updated today, it seems the issue has not been resolved, at least for my account.

[etraut-openai]

Please refer to my post here for reasons why you may be seeing increased usage.

[nishikawa7863]

Not a duplicate. I've reviewed etraut's post on the closed thread and none of the listed factors apply to my case.

I'm on a Pro account. Over the past couple of days, my weekly limit has been draining dramatically faster despite the same workload, regardless of model (GPT-5.3 or 5.4), reasoning effort, or channel (Codex App, CLI, VS Code). No sub-agents, no experimental features enabled. I've burned through ~70% of my weekly limit in a single day on what I'd consider normal, moderate usage. This is a clear regression from last week's consumption rates.

The regional sync fix mentioned in the closed thread doesn't appear to explain this. The drain is consistent and directional, not oscillating.

[cy-ooi88]

Not a duplicate. I've reviewed etraut's post on the closed thread and none of the listed factors apply to my case.

I'm on a Pro account. Over the past couple of days, my weekly limit has been draining dramatically faster despite the same workload, regardless of model (GPT-5.3 or 5.4), reasoning effort, or channel (Codex App, CLI, VS Code). No sub-agents, no experimental features enabled. I've burned through ~70% of my weekly limit in a single day on what I'd consider normal, moderate usage. This is a clear regression from last week's consumption rates.

The regional sync fix mentioned in the closed thread doesn't appear to explain this. The drain is consistent and directional, not oscillating.

Thanks @nishikawa7863, exactly, this is what I am experiencing. I will do a little more evidence / data gathering & post here.

[Xen0m]

Same here never ever burned this much usage on CLI 5.3 codex med.... Lost arround 25% weekly within few prompt on only one project without using fast or anything else.. My 5hours usage goes under 20% which never hapened either.

If this is considered 2 times the normal usage it's straight unusable for any coding activites at least for plus member.

Edit : Just reached 5hours limit for the first time ever.
Token usage: total=742,555 input=697,188 (+ 9,077,504 cached) output=45,367 (reasoning 11,450)
Almost used half of my weekly session on this one.

[VannaGamma]

Please refer to my post here for reasons why you may be seeing increased usage.

I don't really think this has been solved, with all due respect.

First the issue is being attributed to "network", when deflected by actual users the thread gets closed and now "its fixed and thread closed" but we still are experiencing a definitively more than 30% usage attributed to the new model.

That's our opinion based on usage. Thanks.

[bugduino]

Not a duplicate. I've reviewed etraut's post on the closed thread and none of the listed factors apply to my case.

I'm on a Pro account. Over the past couple of days, my weekly limit has been draining dramatically faster despite the same workload, regardless of model (GPT-5.3 or 5.4), reasoning effort, or channel (Codex App, CLI, VS Code). No sub-agents, no experimental features enabled. I've burned through ~70% of my weekly limit in a single day on what I'd consider normal, moderate usage. This is a clear regression from last week's consumption rates.

The regional sync fix mentioned in the closed thread doesn't appear to explain this. The drain is consistent and directional, not oscillating.

I'm experiencing a similar issue on my PRO account and the usage spiked in the last few days despite no changes in my usual workflow

I'm using codex mainly via VSCode extension, current version is 26.311.21342

[sibonyves]

Yeah I was the original poster on that thread @etraut-openai and it's still not fixed. Still using 5.3 codex like I've always been, not particularly high usage compared to previously, but usage limits are still burning very fast.

[calves07]

It's insane how OpenAI insists everything is normal... there are people with Pro subscriptions that are hitting 70% of the weekly limit in 1 day... how is that even normal, specially considering the x2 promo?

Either...
a) the x2 promo is not being applied to everybody
b) there is a bug somewhere
c) the limits were significantly nerfed

And at this rate, Codex will be useless in April considering a lot of people will hit their weekly limits in 2/3 days.

On February/March, I've been often hitting my weekly limits and yet, there were weeks in October when I spent more or the same number of tokens.

[dhpancor]

I completely agree something’s not right. I’ve been using Codex for a few months, and I’ve had enabled subagents since they were released. I usually was ending my week with a 40% remaining usage and using it without worries.

For the past two weeks, I’ve been struggling on my 3rd day and using much less subagents.