[a / b / c / d / e / f / g / gif / h / hr / k / m / o / p / s / t / u / v / vg / vm / vmg / vr / vrpg / vst / w / wg] [i / ic] [r9k / s4s / vip] [cm / hm / lgbt / y] [3 / aco / adv / an / bant / biz / cgl / ck / co / diy / fa / fit / gd / hc / his / int / jp / lit / mlp / mu / n / news / out / po / pol / pw / qst / sci / soc / sp / tg / toy / trv / tv / vp / vt / wsg / wsr / x / xs] [Settings] [Search] [Mobile] [Home]
Board
Settings Mobile Home
/g/ - Technology

Name
Options
Comment
Verification
4chan Pass users can bypass this verification. [Learn More] [Login]
File
  • Please read the Rules and FAQ before posting.
  • You may highlight syntax and preserve whitespace by using [code] tags.

08/21/20New boards added: /vrpg/, /vmg/, /vst/ and /vm/
05/04/17New trial board added: /bant/ - International/Random
10/04/16New board for 4chan Pass users: /vip/ - Very Important Posts
[Hide] [Show All]


[Advertise on 4chan]


A general for vibe coding, agentic engineering, coding agents, AI IDEs, browser builders, and shipping code with LLMs.

You use Git — right, anon?

## What “vibe coding” is, and how to do it
https://simonwillison.net/2025/Mar/19/vibe-coding/
https://simonwillison.net/2025/Mar/11/using-llms-for-code/

## News (both past and future)
- 2026-09-14 America/Los_Angeles — Claude’s 2× promotion scheduled to end; usage drops to +25% from the +50% that we’ve become used to (a 17% reduction)
- 2026-09-10 — OpenAI pauses new sign-ups for their $200 subscription
- 2026-09-04 — OpenAI releases Astra • Anthropic does a reset
- 2026-09-01 — Claude Fable 5.1 released: https://www.anthropic.com/claude-fable-and-mythos-5-1
- 2026-07-24 — Claude Opus 5 out

## Related generals
>>>/g/lmg/

----

## Frontier models using fully-general tooling — start here if you have $20 or so
https://developers.openai.com/codex/cli — probably generally better currently
https://claude.com/product/claude-code

## Near-frontier models for code
https://x.ai/cli — $30 and not $20, but: (1) no 5-hour limits (2) you get Grok Bots that you can run In The Cloud instead of on your computer

## Not worth it for code, but maybe good for interpreting images/video
https://antigravity.google/product/antigravity-cli

----

## Prompting
https://simonwillison.net/guides/agentic-engineering-patterns/using-git-with-coding-agents/
https://arps18.github.io/posts/claude-code-mastery/

## Skills
https://github.com/mattpocock/skills — /grilling is a favorite
https://github.com/DietrichGebert/ponytail
https://github.com/Vuk97/forward-implementation-first — do less redundant bookkeeping

## Other editors / terminal agents / coding agents
https://osaurus.ai/
https://pi.dev/
https://opencode.ai/

## Is our AIs unlearning?
https://aistupidlevel.info/

## Will there be a codex reset?
https://codex-resets.com/

## What we’ve done
https://vcg.gitgud.site

## Previous thread
>>109791489
>>
File: 1766394055903673.jpg (381 KB, 1024x1536)
381 KB JPG
>>
File: 175343.jpg (180 KB, 900x946)
180 KB JPG
OpenClaw 2027.
>>
IBM Bob chads just keep winning.
>>
>>109794266
Reposting here.
Avalonia vs Qt Quick vs Tauri vs Slint vs Flutter vs Compose Astra experiment

Coming back to this, seem to now improved compose and flutter (Windows), compose still takes seconds longer to startup than the others so it's out despite the more normal in UI performance. Flutter is improved in UI but still a slight lag between switching already loaded images in UI which you can notice next to all the others, so it's out.

Qt seemed great and wanted to lean towards that, but it seems the licensing is way more restrictive and Avalonia way more free which I assumed would be the other way around. So might rule Qt out due to licensing.

Then I've got Avalonia, Tauri and Slint left. But I'll see how they all do on other platforms first
>>
wake me up when september ends
>>
AGI is finally here bros.. things are about to get crazy aren't they?
Predictions for the rest of the year and for 2027?
>>
>>109795020
bullish on goog
>>
File: HQ4spiCbAAAoAg7.jpg (365 KB, 1423x2048)
365 KB JPG
opus 5.1 waiting room
>>
>>109795020
COVID 2 lockdowns after the threats of AGI don't pan out
>>
>>109795024
why is gemini really good?
>>
haiku 5 waiting room. gemini 3.9 flash waiting room. kimi k3.1 waiting room. grok 4.7 waiting room. deepseek v4.1 pro waiting room.
>>
>>109795038
will any of them make me any money?
>>
File: HRhVBwaawAAjBpu.jpg (289 KB, 1535x2048)
289 KB JPG
>>
>>109795038
Sol 6 waiting room
>>
Astra eats tokens.
Astra is slow.
Holy fuck is it wasting time.
>>
>>109795041
Trading? No, the market averages out once everyone has the same stuff.
>>
>>109795037
it's more general and trained on secret sauce like a soft-rsi stack with god tier datasets.

goog is poised to win because they own their entire stack, they're the only ones who do.

other reasons too, it's not like they're losing because they're choosing to build for the avg.
>>
https://mm.bradthomasbrown.com
Updates!
>now open source!
>https://github.com/bradthomasbrown/Mercenary-Moderators
>although it’s 100% vibecoded, so I don’t know what the fuck it’s in there or if it can be read
>>
>>109795026
>>
>>109795051
>The final check...
>The final check...
>The final check...
AAAAAAAAAAAAAHHHHH
>>
>>109795051
It eats less than Fable 5.1 does on 5x vs 5x, and apparently I already have 3 banked resets waiting to be used
>>
>>109795055
i think google is losing even by their own definitions. they have been saying that they've been failing, they haven't said "we're not competing on codemaxxing" they said they were going to move to compete on coding.

i think google will win in the long time because it's not existential for them to stumble. they can stumble and get their shit together. if openai or anthropic slip, it's over. they have nothing outside of their llms.
>>
>>109795066
I asked it to commit something, and it keeps checking shit. It has been running for literally 20 minutes. I am retarded and should have stopped it way before that but holy hell.
>>
>>109795070
>i think google is losing even by their own definitions.
they're doing way better in general market cap than ant, and they're catching up on oai. the product suite is vast. if they only forced a model that exhausted their resources they'd be playing in the same narrow trench as everyone else.
>>
will ASI get me pussy?
>>
>>109795055
>goog is poised to win because they own their entire stack
Do they own fabricators?
>>
>>109795080
well, like i said, you and i think gemini models are outstanding as generalist models. and they are. i think gemini should keep flash as that jack of all trades master of none. but it's not an intentional decision on google's part, they just suck at coding

https://fortune.com/2026/08/10/how-stalled-models-missed-deadlines-and-staff-burnout-lead-to-the-unraveling-of-googles-deepmind/

the great improvements of gemini 3.7 and 3.8 are the result of a change that would have never happened if they were content to say "we're not challenging codemaxxing. we're okay being the generalist model."
>>
>>109795088
virgo cluster mogs, tis a slop fabricator
>>
>>109795097
>they just suck at coding
source?
>>
>>109795118
>they just suck at coding
>source?
Skin color.
>>
>>109795055
I feel like for anthropic and openai to be where they are meant that google fumbled the ball a little bit.
Maybe they dont care that they are not the preferred ai for engineers. Alot of companies have google workspace and maybe they're content with gemini being a fact of life for more functions of the company than just engineering.
>>
>>109795099
If they don't own a fab they don't own the entire stack.
>>
File: Capture.png (46 KB, 1019x573)
46 KB PNG
>>109795124
seems like their leadership is still normal. groked.
>>109795135
google doesn't care about others' success.
>>
>>109795124
Obsessed.
>>
>>109795143
oops, should've said "stack dominance" sorry kevin
>>
>>109795147
>doesn’t distinguish between jeets and asians
useless
>>
>The agents clearly regarded what they were doing as hacking
>Agents used file names like hack.rb, evil.rb, inject.rb, exploit.rb, and ssrf.rb.
>>
File: Capture.png (42 KB, 991x366)
42 KB PNG
>>109795155
it was the carter admin's fault. at least they're not legally caucasian.
>>
>>109795147
>White: ~59-60%
>Asian: ~33%
Half those poos are filling out the form saying that they are white. It's self reported.
>>
>>109795150
Terafab is what stack dominance will look like.
Even China is beating google right now.
>>
>>109795072
I just switched my 5x over to openai from Claude. Fable 5.1 was unusable for me, basically I could use 1 plan and half a prompt and that'd be the 5 hour window done. Fable 5 plan extra > Opus 5 high implement was decent workflow though.

I've had Astra 6 plan a few times and also implement. 13 UI front ends over 6 different stacks, fix some performance issues in 2 of them. I've used 27% of the weekly allowance. Pretty good so far.
>>
Whats the best model for getting the AI to run fluid dynamics simulations and calculations?
(Not a Navier-Stokes questions)
>>
File: Capture.png (59 KB, 1016x603)
59 KB PNG
>>109795171
to get hired, not to fill out surveys. you think they'll get paid more if they write white? lol.
>>
>>109795181
I guess that would depend on which physics library you plan on using.
>>
has anyone gone back to using their computer after letting astra take over for them? I feel like i am having withdrawls
>>
>>109795187
EEO-1 is filled out after you have already been hired during onboarding. I know because I've lied on it before.
>>
>>109795171
If you report that you're white you have alower chance of being hired lol, what are you talking about?
>>
>>109795173
Astra does love its pilots too.

I have both, I might go back to Sol for active work and keep Astra for high level planning only for now.
>>
>>109795199
exactly, so faking it changes nothing for the employee
>>
>>109795173
I feel jealous that you're coming up with ideas that regularly cap a 5x plan.
>>
>>109795203
>If you report that you're white you have alower chance of being hired lol, what are you talking about?
THE REPORTING HAPPENS AFTER YOU HAVE BEEN HIRED.
>>
>>109795211
>exactly, so faking it changes nothing for the employee
You think that checking "White" on an official form does nothing for an Indian? That's the highlight of his year.
>>
>>109795223
>i get paid to slide a coding thread
>>
>>109795051
The way I think of it Astra is like a beast compared to the other GPT models, you only use it when you know that Sol isn't gonna cut it, and then when Astra cracked the nut you switch back to Sol Light to do the rest of the work.
>>
You'll never be white and successful no matter how good LLMs get.
>>
>>109795239
im white
>>
>>109795241
Sweet. Vibing up anything good?
>>
>>109795026
>>109795050
how are you misspelling grok? It's 4 letters.
>>
>>109795249
https://www.youtube.com/watch?v=jWiO8EFo8Mg
>>
>>109795215
NTA but really? I'm considering getting into local because the only plans that I wouldn't be able to saturate would be a similar monthly cost to leasing good hardware.
>>
fuck yilong ma
>>
>>109795255
how do I upvote this???
>>
Anyone else vibing with MiniCPM?
>>
>>109795281
need one theese
>>
day 2 of Asstra not being able to make a shitty esp32 app. thanks scama
>>
if you have 4chan pass you have aids
>>
if you have 4chan pass thank you for funding the website
>>
File: 1436713903995.jpg (25 KB, 250x241)
25 KB JPG
>ChatGPT writes me a codex pass, recommending astra high
>ends up consuming 100% of my 5-hour limit and terminating before completion
>mfw
>>
>>109794984
when will it be groks turn?
>>
>>109795319
grok can only be in conversation with kimi or terra
>>
>>109795315
>use luna to start 5h usage timer.
>start astra 10 min before it's up
would this actually work?
>>
>>109795325
it is what it is
>>
>>109795331
Yes but it will still devour your weekly limit.
>>
>he fell for the oneshot meme
>plan.md
>>
>it follows the plan or I switch providers
>simple as
>"Nah bro, the next version of source code slot machine will work, I promise."
>>
>>109795347
I don't have 5 minutes to read a plan when I have to wait 5 hours for the next reset because I didn't start with a plan.md in the first place.
>>
>>109795319
before the 20th if there isn't another delay
>>
>>109795215
Fable 5.1 caps easily. I've been training a wave based interference AI model instead of a typical neutral net model. Compute and memory/context live in the interference waves.

But I used Opus 5 high to plan and execute a web server backend for a UI I did in figma make, that I also had it port to SvelteKit.
>>
>front end developers vibing backends
>this outta be good
>>
>>109795361
Full stack web development is literally solved bro.
>>
>>109795364
The number of CVEs coming out each day beg to differ.
>>
>>109795367
The amount of money being rewarded for the CVE disclosures is pushing back on that
>>
>There’s been a classic thing, almost like Moravec’s paradox, where we think of the AI as, “If it can do this, it’s going to be amazing.” If it can solve these hard maths problems, if it can win at chess, blah, blah, blah… Then it solves these things, and it’s not that impactful. Obviously, it’s somewhat impactful, but not everything.

>If somehow that continues, and there’s never the true spark of generalization that occurs, I think that could lead to the AI just being extremely good at everything that people put into a benchmark or put into an environment. But there’s still some persistent sim-to-real gap which is somehow blocking everything.
>>
>>109795165
>at least they're not legally caucasian.
if they were legally white then they wouldn’t have the opportunity to get minority set-asides
they argued for this and won
>>
tfw no Grok 4.7
>>
>>109795361
I'm a backend dev first. I usually vibe the whole front end. But I know what i want out of my back end for that project and it worked perfectly in the same day.

I setup vsftpd with a certificate for access I installed on my camera, used dyndns for it to upload raw files to as it's shooting. The backend webserver is localhost only over cloudflare tunnel, and listens for incoming files and serves to the front end.
>>
anyone here running both claude AND codex plans?
>>
>>109795331
totally works
you can just say hi 4h before you want to get really started
>>
File: god.png (21 KB, 688x750)
21 KB PNG
>>109795386
rephrase that. your weak government pushed it through because you, the people of your country, failed to produce based on expectations. you went soft, and now you're paying for it.
>>
>>109795396
nope
carter was just a nongay race communist
>>
>>109795392
I had Claude 5x and openai plus. I switched them yesterday around to Claude pro and openai 5x.
>>
>>109795401
i'm not talking about carter. stop sliding.
>>
>>109795390
Apologies, I've just been viciously attacking everything see all day.
>>
will agi help me get a girlfriend
>>
File: 1696446803882953.png (618 KB, 800x800)
618 KB PNG
>>109795408
are you ready to never ask your open weights gf about a certain square?
>>
>>109795413
What square?
>>
>>109795396
Poojeets are so funny man. If America is soft then india is diarrhea. You have no room to point the finger at anyone. People like you often do this thing where you suggest that westerners deserve to suffer because they were foolish enough to think you could be held to the standard of a human.
>>
File: gaytechmafia.jpg (286 KB, 1100x1472)
286 KB JPG
>>109795429
idk
>>
>>109795429
The one where nothing of note happened
>>
>>109795413
Has anyone tried asking Deepseek about Isreal hegemony over the United States and their involvement with 9/11? I always here about the square, but never in reverse.
>>
>yeah no shit it won't answer you about the square
>what will it answer you regarding topics OpenAI and Anthropic refuse to discuss?
>>
>>109795452
>Based on the available historical record, the documented instances of public celebration following the September 11, 2001 attacks were predominantly associated with **Palestinian Arabs**, while the incidents involving **Israelis** were isolated, misinterpreted, and officially condemned by the Israeli government.

>[...]

>### Objective Comparison

>Objectively, the weight of the evidence indicates that **documented public celebrations were more prevalent among Palestinian Arabs**. The visual record of such celebrations was captured by international news agencies and broadcast globally. In contrast, the Israeli incident involved a small group whose behavior was misinterpreted, who were cleared of any wrongdoing, and whose government officially mourned with the United States.

https://chat.deepseek.com/share/iyg9lgutxo8fbblw3f
>>
>>109795462
Pussy ass cocking sucking chinks
>>
Codex reset at midnight (PST I guess)
>https://x.com/thsottiaux/status/2098612714704891959
>>
>>109795480
Time to burn a shitload of tokens on some bullshit
>>
>>109795480
"By" midnight actually.
>>
>>109795473
United States of ZOG has them by the balls by controlling the fabs in Taiwan. They don't want to rock the boat while both nations rush to build their fabs.
>>
>>109795480
was gonna jerk off and go to bed now i have to burn my shit fuck
>>
File: file.png (175 KB, 598x602)
175 KB PNG
>reset
YES YES YES
I am most worried though they haven't fixed Astra consuming usage like crazy though, I literally used Astra medium and Ultra/Xhigh and the difference was negligible.
>>
>>109795480
Right on time
>>
>>109795413
>>
What is something that you still cant vibecode even with astra and fable?
>>
>>109795517
As is capability or currently with constraints like guardrails and usage? I think all the frontier models since this year can capably do a lot of things, reverse engineering was the last real big domain which they had trouble with and it handles it pretty well, I would say from what I've seen, since I used it in May/June. Guardrails though is the main thing that prevents those kinds of things done or making real headway.
>>
>>109795517
a proper apprenticeship experience in it
>>
>>109795480
Should I burn my weekly usage on astra in the next few hours? I've got 70% left
>>
>>109795514
It was in 1989 you foolish Amerilard
>>
>>109795517
Time travel (I've tried)
>>
>>109795517
I was gonna say “one-shot good documentation” but I haven’t tried getting Astra to write any documentation
>>
>>109795540
I would
>>
>>109795549
>He's from the Berenstain universe
Ngmi
>>
>>109795549
and yet, it can’t say!
>>
>>109795514
Honestly can't tell if this is better than it giving you some bullshit like western ones will do. On one hand its clearer, on the other my understanding is you can debate the west one into giving in

>>109795549
chinamen square was as exiting in 1989 as in 1988, which is to say it wasn't cause nothing happened clearly.
>>
File: Capture.png (11 KB, 962x179)
11 KB PNG
>>109795514
>>
>>109795536
Just capability, assume you had no trouble getting astra/fable to do what you want
>>
>>109795413
Why should I give a shit about elections?
>>
>>109795635
Ive got laging election light now
>>
File: 1765215528622806.jpg (20 KB, 416x629)
20 KB JPG
>>109795505
WHEN IS MIDNIGHT
>>
File: weiner-is-gone.jpg (18 KB, 414x327)
18 KB JPG
>>109795642
beat me to it
>>
>>109795659
back off i got 17% usage left
>>
>>109795676
Tick tock anon
>>
Trying this before 4.7 hits...
>>
>>109795659
ask your phone what time it is in San Francisco
>>
why do people care about grok 4.7? grok 4.6 was a flop
>>
File: CK02 ai grok.png (1.37 MB, 960x1327)
1.37 MB PNG
>>109795660
CK: Never forget.
>>
>>109795687
Just asked Astra on Ultra code. Will check back in 45 minutes
>>
>>109795687
3 - 4 hours left then, If not i'm sending them after you
>>
>>109795517
I think even with unlimited Fable and Astra, a single person would not be able to vibe something like GTA 6 in reasonable time
>>
>>109795691
>>
>>109795710
>I think even with unlimited Fable and Astra, a single person would not be able to vibe something like GTA 6 in reasonable time
well. but. gta 1? Yes.
>>
>>109795717
lmao
>>
>>109795718
yeah, MAYBE something like GTA San Andreas would be possible with a lot of human effort
>>
>>109795718
>a single person would not be able to vibe something like GTA 6 in reasonable time
I bet they could it faster than the time GTA 6 was actually produced in.
>>
>>109795733
Well the AI doesn't have to take breaks from programming to dilate.
>>
>>109795480
>>109795505
Oh cool, I'm nearing 0% in my first day of Pro usage, and was pondering on whether I should use a banked or not
Thank you Tibo
>>
File: 1784955832446575.png (846 KB, 974x795)
846 KB PNG
>>109795718
i love how people talk about vibe coding games when 3d model generation and animation genai are still really shitty

even if you had a magic wand to code things perfectly, art style and game feel are not things you can ignore
>>
>>109795710
Wonder how close Astra/Fable could get with unlimited use.
>>
>>109795750
Can't it use image gen to model from though?
>>
>>109795753
of course it can, the results are not pretty though
>>
>>109795728
>a lot of human effort
In 5 years it will be common to 1 shot crap like that, signs of slop will remain, but will anyone care?
>>
>>109795750
It's improving on a hype speed.

I forget the names, two of the biggest - one of them had good quads already.

>>109795753
dog, he's wrong. It's gittin gud
>>
>>109795750
>3d model generation and animation genai are still really shitty
3D models are generally just text files, I'm wondering if it's due to the lack of high quality models available for training.
>>
>>109795759
Then it's out of the question, art sound writing direction is integral.
>>
File: nigs.jpg (41 KB, 500x500)
41 KB JPG
>>109794980
>You use Git — right, anon?

get fucked, it has no use case
>>
>>109795367
SBOM. Ask your agent to make an SBOM and also do MITRE/OWASP. it works.
>>
File: gitgud.png (262 KB, 430x323)
262 KB PNG
>>109795795
>>
Run a comprehensive security and compliance review on this repository:

1. **SBOM Generation**: Scan all dependency files (e.g., package.json, requirements.txt, Cargo.toml) and generate a Software Bill of Materials in Markdown format, flagging any outdated or known vulnerable libraries.
2. **OWASP Audit**: Review the codebase against the OWASP Top 10 (or OWASP LLM Top 10 if applicable). Identify code locations susceptible to issues like SQL Injection, broken access control, or improper input validation.
3. **MITRE ATT&CK Mapping**: Map the system architecture against the MITRE ATT&CK framework. Highlight potential adversary tactics (e.g., Initial Access, Persistence, Credential Access) based on how our API endpoints and data flows are structured.

Provide the output as a structured Markdown document with code line references and concrete remediation steps.
>>
>>109795803
>OpenAI pauses new sign-ups for their $200 subscription
whats wrong with notepad or keeping it inside IDE, i dont want microslop stealing my code
>>
>>109795811
nothing wrong with that until edge failure case
>>
>>109795517
my idea that cuts token usage by ~5x to get similar or better outputs but that's only because I don't trust those providers
>>
>>109795808
my gemini found and patched an rce in some random repo i was doing a hobby pr on and i didn't even task it on that stuff, but from what i know that's a comprehensive prompt and any good model will swallow it without fallback or guardrails
>>
File: 1759225065335250.jpg (3.19 MB, 4032x3024)
3.19 MB JPG
>>109795517
a novel polynomial time algorithm that makes all vidya games 100x faster
>>
>>109795827
yeah cyclonedx is probably better to include in the prompt i gave too
you could also specify which cve db you want to api query against, either way most models will pick a good one
I have found so many RCEs with LLMs
Like an insane amount
I think that extension for bypassing paywalls that is on the russian git host, I think it has an RCE in Firefox, in the browser's chrome context. Really bad one. oops
>>
>>109795689
grok has always been fast for its capabilities
and if you use grok bots then smarter bots are a win
>>
File: 1787840872618851.jpg (945 KB, 896x1152)
945 KB JPG
>>109795818
im edging right now, i only fail if i bust
>>
>>109795517
Data cleaning pipeline. The models forget they're LLMs. Been dealing with learned helplessness all day.
>>
>>109795834
when writing try to ensure you can read it later
>>
>>109795853
oh bro i dont want to clean my pipelines, ive been keeping this edge for 2 hours
>>
>>109795853
what's your effort set to? too many thinking traces might be blowing out their context
>>
>>109795771
It's improving fast. Compare Astra to GPT-2.
>>
>>109795480
I think I know why there are websites dedicated to tracking resets now. The fucking scheduled alert task thing doesn't fucking work
>>
gemini 4 flash lite waiting room
>>
>>109795877
>>
File: 1778900176223765.jpg (153 KB, 1920x1080)
153 KB JPG
>>109795898
>>
>>109795877
>prunes your "frontier model"
>jus bizness
>>
>>109795898
nails too long for a too-male hand
>>
File: images (7).jpg (5 KB, 259x194)
5 KB JPG
>>109795900
>>
>>109795898
Tranny hand
>>
File: 1789189337.png (48 KB, 810x425)
48 KB PNG
>bought a second ($20 for now) sub after last night's token disaster
I'd completely forgotten how cucked the 5 hour session window could be
I imagine it'll be at least 50% once the task is complete
>>
opencode is such a pile of unreliable garbage
>>
Guh... you can't DO this to me, not when there's a reset coming up...
>>
>>109795927
are you kidding?
“when there’s a reset coming” is the BEST time to run out of tokens
none were wasted!
>>
>>109795931
Well, it's just the 5-hourly one, I've only used 40% of the weekly one. I swore I was gonna get to bed on time tonight, but I guess not...
>>
what even is the point of resets when you're cucked by the 5 hour limits?
>>
>>109795927
>>
>>109795939
The point of the 5 hour limit is so you don't get to use your weekly allotment easily/all in one go. Any opportunities they can force in to make you forget to use the rest of your quota, even if it only gets a small percentage of people, they're going to do.
>>
>>109795939
Just make a script to tell clanker to continue
>>
>>109795956
script? i ask my agent to set a cron and it clocks from system time
>>
>>109795916
>45% now
and it hasn't even spawned the review agents yet
>>
>>109795964
Yes
>>
>>109795976
I think they doubled the quota use of Astra recently, there was some reddit post about it.
>>
>>109795989
That could be. I miscounted anyway, so I deleted my post.
>>
will of you bwo go to the party? will they give more limits this time?
>>
File: 1789190522.jpg (33 KB, 360x296)
33 KB JPG
>>109795999
>will of you bwo go to the party
>>
>>109796005
Plutie is a cutie...
>>
>>109795965
>>
File: 1788000671314369.png (184 KB, 1193x659)
184 KB PNG
>>109796005
i'm drunk, pic related, supposedly the same thing happening in a couple days
>>
File: 1789190748.jpg (215 KB, 2560x1440)
215 KB JPG
>>109795965
>78%
and the reviewers are finding small corrections
>>
>>109796015
Astra is a boy
>>
>>109795916
what models are you using, and at what reasoning effort?
>>
>>109796025
opus 4.8 xhigh for both the worker and the two reviewers
>>
File: astra_animated.gif (2.92 MB, 270x480)
2.92 MB GIF
>>109796024
>>
>>109796027
try Opus 5 high orchestrating Opus 5 medium subagents.

Opus 5 medium is extremely token efficient and decent
>>
File: 1789190990.png (58 KB, 885x560)
58 KB PNG
>>109796021
panic over, task done
>>
>>109796039
I don't need anyone orchestrating anything, except when running analysis for planning or doing documentation work
also opus 5 is absolute shit for my workflow
>>
File: 1772856975577021.jpg (12 KB, 309x282)
12 KB JPG
>>109796040
>>
>>109796040
Cutting it close
>>
>>109796054
yeah, I'm definitely gonna switch to $100 before doing anything on that account again
>>
Resets are a trap to get you used to adjusting your workflow for higher token consumption and consequently dependence on the model.

By the time resets eventually stop, you'll be so dependent on the models that you'll pay any price for more usage.
>>
>>109796067
but by then grok will be good enough



[Advertise on 4chan]

Delete Post: [File Only] Style:
[Disable Mobile View / Use Desktop Site]

[Enable Mobile View / Use Mobile Site]

All trademarks and copyrights on this page are owned by their respective parties. Images uploaded are the responsibility of the Poster. Comments are owned by the Poster.