Curious🧐

science lover, likes to play cricket,curious at spirituality, knowledge seeker🏏🔭💡🕉️

[Jack @VKx76]

Watch this. Your perspective will change.

Most Indians believe Mahabharata happened.

But very few can prove it.
Chandru Ramesh can.

A former software engineer who gave up his corporate career to spend years travelling the world, collecting physical evidence, inscriptions, archaeological findings and over 300 data points backed by 1,000+ citations.

Mahabharata and Ramayana are not just stories we were told…
They are our History.

[Curious🧐 @curious_nerd1]

@arsh_goyal RL

[Darren @dtzy_88]

才发现 Codex「额度用完还能跑完任务」的福利前几天还有的，结果被两位傻逼搞没了🤬

时间线 👇

8/20：OpenAI 员工 Angel 发帖：额度打到 100%，
也不会掐断正在跑的任务，是 OpenAI 刻意设计的，“在别人干活干到一半时停掉，没什么帮助”。这时福利还在

8/21：网红 Theo 贴脸开大，转发并当场示范怎么滥用：开 /goal 跑长任务，额度归 0% 之后，又烧掉至少 $800 的 tokens

同一天，Theo 智商 250 的实习生 Maria 直接把这招做成了一个叫 Endless 的 harness。

原理是这样的：
- 开一个 turn，给 agent 塞个 wait_for_user_input 工具
- 任务做完不结束 turn，而是调用这个工具等待
- 你的下一句话当工具结果喂回去

整个会话永远锁在同一个 turn 里，quota 检查永远轮不到触发

宣传语原话：「infinite inference (kinda)」，榨干你订阅的每一滴

8/22：Endless 帖子爆了，一堆人拿它在归零后继续白嫖，眼看演变成大规模滥用

Maria 当天道歉：“没想到这个小点子危害这么大，如果导致大规模滥用我愿担责”
当场下架 npm 包、关掉网站、删了 GitHub 仓库

最骚的还在后面：
删完库，她转头喊话 Codex 团队：“希望你们排查一下，把这些账号封了，这样我们就安全了 😊”

不是，你傻逼吧，居然呼吁 Codex 团队封掉用你工具的账户？？？哪有人那么背刺信任你的人的 💩

这一轮风波的结果就是：现在 Codex 额度用完，任务立马停

反观：
Theo 悄悄删了那条 $800 示范帖
Maria 的 Endless 下了，帖子也删了

点火的全身而退，火上加油的道了歉，最惨的是连 Endless 都没听过的我们

我们的 「额度用完还能跑完任务」福利是 TMD 真的没了啊，气死了 😤

https://t.co/byTyLfOnJW

[Curious🧐 @curious_nerd1]

@AdiShudhalwar @yoursbidhuri Then debunk it

[Curious🧐 @curious_nerd1]

@thsottiaux
I am ready to share the data. It is ridiculous that I had just given a single prompt of reading and updating a document which is of just 800 lines and my usage just dropped to 80% from 100. Seriously, there is something wrong with the codex. Reset is not soultion

[Osvaldo Chávez @imosvaldoe]

Update on the token quota issue @thsottiaux

I decided to run a controlled test using GPT-5.3 Codex Spark.

I started with 100% of my 5-hour Spark quota and 97% of my weekly Spark quota. I asked one question: "Audit the tokens I've used today."

The audit reported 679,010 total tokens, but 619,008 were cached input tokens. The actual new output was only a few thousand tokens.

After that single prompt:

• 5-hour Spark quota: 100% → 96%
• Weekly Spark quota: 97% → 95%

Something still doesn't add up.

I'm not claiming this is a bug. I'm asking for transparency. Right now, it's impossible to understand what is actually being counted toward our quotas.

This is incredible.

[Curious🧐 @curious_nerd1]

@thsottiaux I am already down to 50 percent

[Curious🧐 @curious_nerd1]

@thsottiaux No, I just used one prompt. And my my usage is down to 90 percent from 100

[Curious🧐 @curious_nerd1]

@Hlamyan31 @RangasthalamIN What is spreading fakes news?

[Akshay 🚀 @akshay_pachaar]

Sam Altman made the case for open-source harnesses in July.

a month later, someone shipped it, and it's more efficient than most managed harnesses.

here is the problem it was aimed at:

a large share of your agent's token bill is the model rereading things it already read.

that isn't the model's doing. the runtime around it decides what goes into every prompt and how often the model gets called.

for example, an agent queries a CRM at step four and gets back 400 rows. those rows get piled up in the conversation history.

by step nineteen, the model has to read those rows fifteen times unnecessarily, and every token read is billed at input rates.

it happened because your harness assembled that prompt on every turn and kept the rows in it.

that gives you two levers: how much context the harness carries forward, and how often it calls the model.

there are four practical ways to keep the prompt from growing unnecessarily:

→ load tool schemas on demand. a server with 100 tools doesn't need to put all 100 into every prompt when the agent only calls two.

→ offload large results to disk. turn a large response into a short preview and a file path instead of replaying the entire result on every turn.

→ delegate to subagents. let a subagent spend thirty tool calls in its own context and return one summary to the root agent.

→ run toolchains in code. one script calls three tools, joins the results, and returns a table instead of three turns each dragging a full response.

but reducing context is only half the job. you also need to control how often the model gets called.

a good harness should avoid unnecessary planning, verification, and reflection when the work can be completed in fewer steps.

@TrueFoundry's open-source agent harness, TrueForge, is built around both of those controls.

it sits between the model and the tools, deciding what goes into every prompt and when another model call is actually needed. it also breaks token usage down across the harness, skills, instructions, tools, and messages.

DevRev's Enterprise-Bench is where this gets tested, on multi-step tasks of the kind where an agent pulls records from one system and reconciles them against another.

TrueFoundry ran TrueForge there against Claude Managed Agents, both on the same model, and both finished the same number of tasks.

the tie is the part that matters, because it means the gap underneath is not a quality tradeoff.

TrueForge reached that score on close to a third of the tokens, with roughly 40% fewer trips back to the model. for the same result, that comes out around 2.7x cheaper than Claude Managed Agents.

swapping in an open model made it sharper still. TrueForge with GLM-5.2 scored a little higher than either setup above, and the entire benchmark run cost about $3 at list prices.

being open source matters beyond the license here. the model underneath can be swapped without rewriting the agent, and the whole thing can run inside your own environment when the data cannot leave it.

all of this comes down to the runtime around the model, the context it carries, the tools it exposes, and how many times it goes back to the model.

that is what a production harness actually owns.

the full task list, the per-run numbers, and the MIT-licensed code are on GitHub: https://t.co/Ueo1InZH7M

(don't forget to star 🌟)

you can read more about the same in the article quoted below.

thanks to the TrueForge team for working with me on this one.

[Curious🧐 @curious_nerd1]

@Robertg761_ Sake here

[Curious🧐 @curious_nerd1]

@Im_IrushiK Tagging @thsottiaux

[Irushi @Im_IrushiK]

This is actually wild.Codex usage has been draining way faster lately,

Same complex task:

Sol Ultra - 6 hours, 100% → 0%, didn’t finish
Sol Ultra (2nd account) - 5 hours, 100% → 10%, got stuck in a review loop
Fable 5 Ultracode - 3 hours, finished, used only 25%

I’ve done similar tasks with Codex before and never saw usage drain this badly until this week.

This's just my experience, Anyone else noticing this?

[Curious🧐 @curious_nerd1]

@Shekswess Fuck you meesho Tibo, for a sec i thought it is real

[Curious🧐 @curious_nerd1]

@jetfam @thsottiaux @thsottiaux i justed few md files for my architecture, in just one day i burned 70 percentage of usage. With 5.5 this isn't case. With $20 plan i used to get done atleast 4 of days

[Curious🧐 @curious_nerd1]

@shithitdfan @koratala_fan @narendramodi Wallet payments and there are only few banks that support zelle

[Treeni @treeni]

> "Have trapped seven school girls — two of them have already Kalma"
> "The moment our time comes, we will enter their (Hindu) homes and finish them."
> "Our aligiance is only with Ummah — our Pakistan."
> "Let our government come.. We will show you.."

YouTuber Fire Ninja exposes Indian Muslims' mindset with a quick social experiment on Independence Day.

[Vishal Singh 🥑 @vishalsingh2972]

Sarvam AI opened Voice Agents today, so I built a Hyderabadi sabzi aunty 😭

Called her and bargained in Hindi + Telugu + English, kept interrupting, switched languages mid-sentence, and she still remembered my order, gave me a final total, and even confirmed my (fake) UPI payment.

Took me ~15 mins to build.

It was fun to recreate the lost art of vegetable bargaining, but this time with an AI agent.

[Curious🧐 @curious_nerd1]

Those who are not in X. Might me thinking there bug in codex that it always shows 100 percent usage left

[Curious🧐 @curious_nerd1]

@kunalpurohit Wow they should have written in sqcm