The thread has little explanation as to what weird thing they’re doing to Codex that is making the default work poorly, and it kind of seems like it’s getting confused about whether it wants to set the caching mode or the breakpoint or both.
In any case, I find the behavior change interesting. It sounds to be like 5.5 and below may have been using a conventional attention scheme where a cached KV sequence can be easily used to restore a prefix of itself, but perhaps 5.6 is using linear attention or LSTM or another recurrent scheme where you cannot rewind the model state by just truncating it.
fakwandi_priv 43 minutes ago [-]
I have noticed the same thing starting with 5.6 when editing my last prompt inside the vscode codex plugin, I’ve seen the model’s thinking respond to the edit with a remark.
Slightly bummed out about it because in the past you could try different situations during a planning session and it wouldn’t pollute the cache but now it does. I’m not sure if forking the conversation has the same problem.
xiphias2 3 hours ago [-]
It’s really cool that we have this proof that US companies are half year behind Chinese models in architecture.
geysersam 1 hours ago [-]
What is the proof?
cma 46 minutes ago [-]
Nemotron was using hybrid with recurrence via mamba layers since around April 2025.
zuzululu 2 hours ago [-]
i dont know what the hell is going on lately i pop in to issues or discussions and its agents talking to each other or telling me what PR to merge
I mean i use AI too but was taken back when an agent popped up dictating what i should do and so on....felt weird
debugnik 47 minutes ago [-]
My most awkward experience was a maintainer commenting on my feature request just to prompt a bot to "explain to issue reporter why this is very hard to implement."
It felt like they were trying to avoid me. They could have simply addressed me and given me the explanation they gave to the bot: it would have been simpler for him and more polite. I did in fact reply without waiting for the bot.
squigz 30 minutes ago [-]
It's as though you're talking to someone and they were said to their 'assistant', "Explain this to this person" and walked away. It doesn't really matter what the explanation is, it's just gross.
DrJokepu 2 hours ago [-]
> It sounds to be like 5.5 and below may have been using a conventional attention scheme where a cached KV sequence can be easily used to restore a prefix of itself, but perhaps 5.6 is using linear attention or LSTM or another recurrent scheme where you cannot rewind the model state by just truncating it.
I feel like this is the kind of substantial change to your product that you would need to tell your customers about. It would be simply disrespectful to your customers to not disclose this upfront.
zx8080 2 hours ago [-]
It depends on who they consider the customers. Shareholders and govt are the customers, not users.
Users is the product.
ryanjshaw 1 hours ago [-]
The fact that companies with access to SOTA non-public AI keep having these kinds of dumb bugs in something as simple as a chat app helps validate my disbelief of people claiming AI is ready to replace all software development.
echelon 1 hours ago [-]
AI is ready to replace both jobs and companies.
The companies you see struggling are ripe for disruption.
remoroid 53 minutes ago [-]
So OpenAI the company making the AI and using it inhouse is ripe for disruption? What?
dozerly 28 minutes ago [-]
I want some of what he’s having please
TheP1000 4 hours ago [-]
Our codex on AWS Bedrock read / write cache ratio was less than 5%. Cache writes are very expensive and they were never being used. This results in codex on Bedrock causing ~10x what it should due to no caching and massive writes.
The workaround in issue resolved for me:
web_search = "disabled"
otterley 3 hours ago [-]
If you’ve got a workaround, I’d suggest updating the issue description to have it up top there so similarly impacted users can spot it quickly and benefit.
the_duke 2 hours ago [-]
It's already mentioned in the issue...
otterley 2 hours ago [-]
It’s in a comment halfway down the page. It’s not in the issue description.
If enough comments are added to the discussion, it might end up being collapsed.
chrisweekly 2 hours ago [-]
"causing" -> "costing", right?
yablak 3 hours ago [-]
Way to bury the lede..
bigbuppo 18 minutes ago [-]
Sounds like a path to profitability rather than a bug.
spacedoutman 2 hours ago [-]
Something is wrong with the codex app too, burning usage like crazy lately.
My conspiratorial mind thinks they're doing this deliberately and using the resets to mask things so people can't tell their limits are reduced. The $200 / month plan covers about 2 days of usage for me right now.
ac29 2 hours ago [-]
There haven't been any free resets in the past week, there were 4 in the first half of the month
2 hours ago [-]
zuzululu 2 hours ago [-]
indeed it has anybody know whats going on at openai ??
dgellow 2 hours ago [-]
Maybe preparing for their IPO?
hk1337 3 hours ago [-]
I wonder if it's related to Codex wearing out SSDs.
vee-kay 53 minutes ago [-]
[dead]
moralestapia 1 hours ago [-]
Funny how it is always more charges but never less or no charges.
"Random" accidents that always go against you, too biased to be random.
But don't notice that too much, you might start to see patterns here and there that you're not allowed to, might get you banned from places, etc.
varjag 47 minutes ago [-]
This can be a reporting bias. Noone opens an issue when they were billed too low.
andrewchambers 19 minutes ago [-]
I doubt anyone announces when they have under billed. OpenAI has also done many low price deals and quota resets.
I think you are just biased.
edoceo 3 hours ago [-]
Loaded question: would an openrouter or similar solution caught this before the $BigProblem showed up?
Here are the docs:
https://developers.openai.com/api/docs/guides/prompt-caching...
The thread has little explanation as to what weird thing they’re doing to Codex that is making the default work poorly, and it kind of seems like it’s getting confused about whether it wants to set the caching mode or the breakpoint or both.
In any case, I find the behavior change interesting. It sounds to be like 5.5 and below may have been using a conventional attention scheme where a cached KV sequence can be easily used to restore a prefix of itself, but perhaps 5.6 is using linear attention or LSTM or another recurrent scheme where you cannot rewind the model state by just truncating it.
Slightly bummed out about it because in the past you could try different situations during a planning session and it wouldn’t pollute the cache but now it does. I’m not sure if forking the conversation has the same problem.
I mean i use AI too but was taken back when an agent popped up dictating what i should do and so on....felt weird
It felt like they were trying to avoid me. They could have simply addressed me and given me the explanation they gave to the bot: it would have been simpler for him and more polite. I did in fact reply without waiting for the bot.
I feel like this is the kind of substantial change to your product that you would need to tell your customers about. It would be simply disrespectful to your customers to not disclose this upfront.
Users is the product.
The companies you see struggling are ripe for disruption.
The workaround in issue resolved for me: web_search = "disabled"
If enough comments are added to the discussion, it might end up being collapsed.
My conspiratorial mind thinks they're doing this deliberately and using the resets to mask things so people can't tell their limits are reduced. The $200 / month plan covers about 2 days of usage for me right now.
"Random" accidents that always go against you, too biased to be random.
But don't notice that too much, you might start to see patterns here and there that you're not allowed to, might get you banned from places, etc.
I think you are just biased.