Szpadel

does it increase or decrease? :)

I usually copy paste snippet or function name that is affected and comment it. similar like you would sit next to coworker and tried to say where he/she messed up. agent is able to get location correctly

I know this is change of context, but I have full screen terminal toggled with keyboard shortcut and with middle mouse paste (Linux) I do not have to move cursor anywhere

the compaction does not compact whole context, but keeps last ~20k tokens as is, I believe this helps a lot to model to not get confused what it is doing right now.

it also have soft/soft compaction limit, it tries to compact on turn boundary when possible. with combining with above this can get you about 35% more context (at least it looks like this with the sol)

codex when shell command is executed, will pull output with hard cap at max 30s, so for running compilation it will burn tokens without any benefit.

I have some tasks where agent will have to run some suite that can take over an hour, and codex burns about $20/h just waiting and reasoning every 30s "yep, that's still running". And what is going to happen after compaction, when whole context was just waiting? it will loose the plot and when I'm back it just does completely different thing that I asked it to do.

codex also have a bug, that opening refuses to resolve that adds your last steer after compaction, so imagine that you asked it to cleanup some tmp files or refactor/simplify something. it will do that again and again after each compaction, best case it just burns tokens and figures out, this is already done, or worse do it again and mess up everything and forget about it's task

in my experience pi does much better than codex for compaction and token use. And this is the main reason I switched to it.

migrating to the same compaction and exact tools as codex uses will make it at the same level as codex so what benefit will it have over codex? sure you can customize tui to your liking and add something on top, but the efficiency gains will be gone

sources for easy confirmation:

https://github.com/openai/codex/blob/98d28aab54ed86714901b66...

https://github.com/openai/codex/blob/98d28aab54ed86714901b66...

https://github.com/openai/codex/blob/98d28aab54ed86714901b66...

there seems to be very big misunderstanding about what the "ultra" is, so let me explain it basing on the codex source code:

it's similar to Claude code ultracode.

there is no ultra effort level implemented on the backend. it's just alias in the codex to max effort setting and single line addition to prompt to use subagents proactively. that's all

as far as we know pro models work differently. for once those are backend implementations and they probably run multiple parallel reasonings for any chunk and use some judgement model to pick best version as persistent one. but that's what I believe is most popular guess, because this is openai secret sauce.

there is still no way to use pro models from codex, or at leat so far there is no trace of it anywhere.

big misconception

open source != open weights

open weights model is like... Winamp for example. it's free, you can download it and use it however you like, you could also do some binary patching or dll injections to alter it functionality but it's not enough to develop next version.

the same is with ai models, weights are the binary final artifacts. for development and improvements you need to have training data, pipelines, RL harnesses, etc.

also of you believe Chinese companies will be releasing weights indefinitely, you are not understanding motivations.

Chinese companies spend significant amounts of money to train a model so why they are releasing it for free? they basically provide researchers starting point for developing tooling and optimizations for serving the model in return. and also get some PR. They also do not have to pay for inference of those models that much, as they probably serve them with loss anyways to gain market. they are gov sponsored so money are not issue there, so they try to speedrun their way to what US companies have. And guess what happens when they reach it. they will stop releasing weights and increase pricing or will use them for gov purposes.

now, that should teach him to sell those on black marked instead

I'm mostly joking here, but Microsoft is one of few companies that handle cyber security in a way that really incentive people to not report them.

it's either by downplaying impact and not paying or paying very little or doing other researcher hostile activities.

especially that someone here mentioned some time ago that black market pays about 3x for the same class of vulnerability, so you need fairly high moral standards to go direct way

> but as far as I can tell those credit costs are an exact match for the API token costs listed for those models.

it is only true for USD. for example if you pay in euro, this is actually more expensive. kind of makes no sense, because it translates to $1 = €1

I'm assuming that right now all valves as computer controlled so why we cannot have best of both worlds? cannot we dynamically switch between Otto and Atkinson cycles by just changing valve profiles?

computer could use Otto cycle in case more power is needed in rare situations

huge red flag

> Non-technical teams are now shipping production code

if you vibe code financial systems this cannot mean anything good for your business

> Zero Electron imports remaining in the codebase

and you know that AI wrote all of it with minimal human supervision.

side note: last few days I noticed that vscode stopped leaking memory all over the place. when left idle it was taking all the ram I had + 20gb of swap space

and recently I noticed that I have half of the ram free.

I use insiders build btw, so stable might still not have those improvements

There is one more feature that zram can do: multiple compression levels. I use simple bash script to first use fast compression and after 1h recompress it using much stronger compression.

unfortunately you cannot chain it with any additional layer or offload to disk later on, because recompression breaks idle tracking by setting timestamp to 0 (so it's 1970 again)

https://gist.github.com/Szpadel/9a1960e52121e798a240a9b320ec...

that's the key information to understand the whole thing.

it would be great if that could be in the article in the first place. (I'm assuming you are the author)

counterargument: I always hated writing docs and therefore most of thing that I done at my day job didn't had any and it made using it more difficult for others.

I was also burnt many times where some software docs said one thing and after many hours of debugging I found out that code does something different.

LLMs are so good at creating decent descriptions and keeping them up to date that I believe docs are the number one thing to use them for. yes, you can tell human didn't write them, so what? if they are correct I see no issue at all.

but isn't it what we wanted? we complained so much that LLM uses deprecated or outdated apis instead of current version because they relied so much on what they remembered

I had identical scenario that killed ssd drive. since then I always place laptop with vents on the top

imagine such support ticket:

I used minimax M2 (context it's very unreliable) for installation and it didn't work and my document folder is missing, help

how do you even debug this? imagine you some path or behaviour is changed in new os release and model thinks it knows better? if anything goes wrong who is responsible?

Previously they didn't officially quote how much limits were included in pro subscription, but you could determine it by upgrading from plus that reached weekly limits - after upgrade you ended up with 8% used limits, so we can assume they reduced limits by half just for pro users.

inference costs nothing in comparison to training (you have so many requests in parallel at their scale), for inference they should be profitable even when you drain whole weekly quota every week

but of course they have to pay for training too.

this looks like short sighted money grab (do they need it?), that trade short term profit for trust and customer base (again) as people will cancel their unusable subscriptions.

changing model family when you have instructions tuned for for one of them is tricky and takes long time so people will stick to one of them for some time, but with API pricing you quickly start looking for alternatives and openai gpt-5 family is also fine for coding when you spend some time tuning it.

another pain is switching your agent software, moving from CC to codex is more painful than just picking different model in things like OC, this is plausible argument why they are doing this.

for that you need new socket and motherboard. you need to physically route those extra lanes to pcie slots or other components

>why do the results need to be decrypted by trustees after the election?

they probably design this system to be used for government elections, how they can convince anyone to use it when they do not use it for their own elections?

I gave it a spin with instructions that worked great with gpt-5-codex (5.1 regressed a lot so I do not even compare to it).

Code quality was fine for my very limited tests but I was disappointed with instruction following.

I tried few tricks but I wasn't able to convince it to first present plan before starting implementation.

I have instructions describing that it should first do exploration (where it tried to discover what I want) then plan implementation and then code, but it always jumps directly to code.

this is bug issue for me especially because gemini-cli lacks plan mode like Claude code.

for codex those instructions make plan mode redundant.

what are tested and fairly lightweight alternatives for Loki?

elastic stack is so heavy it's out of question for smaller clusters, loki integration with grafana is nice to have but separate capable dashboard would be also fine

isn't that weird there are no benchmarks included on this release?

from time to time your trusted supplier might be out of stock and you need drivers quickly

even backblaze bought drives in supermarket when there was HDD shortage

how they deal with plain spam? if it's unmoderated and anonymous why it isn't floating in see of bot generated spam?

i think this might be caused by codex. it's open source, many people use it and it uses ratatui. People check how it is implemented and discover ratatui.

I believe this might be current most popular application using this library.

I'm surprised it isn't included in this showcase

Yep, my memory was incorrect here and I didn't had access to computer, but it is true with `0 <= null`

well there is also one weird quirk I I assumed will be also included in this article:

because a <= b is defined as !(a > b)

then:

5 < NaN // false

5 == NaN // false

5 <= NaN // true

Edit: my bad, this does not work with NaN, but you can try `0 <= null`