Akage

Learning

[Akage @The_Akage]

WordCanvas3D is live 🚀

Watch text snap into tokens & swirl into 3-D word galaxies - exactly how GPT “sees” language.
Built with NextJS, Three.js & Python.
Fly through embeddings in your browser, no login required.

Try it → https://t.co/txckdtJ3Es

[Akage @The_Akage]

@Polymarket I mean, human babies do all that when growing up, so astra is definately on right track

[Akage @The_Akage]

@NathanMilner9 @DramaAlert At this point GTA 6 became the second most delayed thing, first being this WW3 😂

[Akage @The_Akage]

@synthwavedd @thsottiaux And this wasnt a long horizon task. it was 3 minutes into execution

[Akage @The_Akage]

@synthwavedd Not just astra but other models as well. Even luna/sol seems to have suddenly gone full blind on context
Even just few hours ago:
it was on right track, then after running cmd it completely went in opposite direction?? did it lose all context in mid execution?? @thsottiaux

[Akage @The_Akage]

@karanC_12 but you know.. google controls majority of internet.. they can just.. loosen their gaurds or give unfair hints to the model in the training. Google already has every companies data

[Akage @The_Akage]

@thsottiaux @stemonteduro How exactly is buying credits a solution to this? 😂
If someone paid 100$ for a service and gets a shitty experience, then you expect them to pay 80$ for more shitty experience?

How does buying credits solve the buggy desktop app or less usage issue?

[Akage @The_Akage]

@tibo @ChatGPT @thsottiaux Tagged the wrong guy by mistake

[Akage @The_Akage]

@tibo @ChatGPT What is happening with codex? Anyone else having major Codex issues lately?

• Usage limits drain much faster, even with Sol/Luna.
• Worst: Codex sometimes works on the current prompt correctly, but the final response answers the PREVIOUS prompt...

[Akage @The_Akage]

Lower limits were one thing, wrong outputs are much worse 😥
It has messed up my daily reports because of these things. Not expected this from openai :(

[Akage @The_Akage]

... Sometimes it can’t even see my latest prompt and repeats work from the previous one or hallucinates stale context.
- desktop app is most buggiest its ever been. Always lags for good 5-10 seconds when launch. Just opening app or chat is enough to consume 1-2% usage limits

[Akage @The_Akage]

Wait we still had some enigma msgs from 1940s unsolved?? 🫪

[MTS @MTSlive]

SITUATION DETECTED: GPT-6 Astra cracked a previously unsolved 1941 German Army Enigma message in about 10 hours, searching archives, writing attack code, and testing keys autonomously.

[Akage @The_Akage]

@PopBase We got a new cat species before GTA 6 🐈

[Akage @The_Akage]

@zane_os Because Deepseek was already popular outside china 🐳

https://t.co/z5FS4weJ5b

[Akage @The_Akage]

Just saw too much of deepseek maid cuteness on my timeline so i had to do it 😂

#deepseek #meme

[Akage @The_Akage]

@yur_o_0 😂

[Akage @The_Akage]

@zarazhangrui Same! Its so obnoxious. my conversation usually goes:

"So does the function you implemented work now?"
-*some claud-ish language non sense*

"I have no what you just said. explain in simple human language."
-*more non sense*

"pls answer in yes/no does it work"
-No, but....

[Akage @The_Akage]

@teortaxesTex

[Akage @The_Akage]

@zane_os

[Akage @The_Akage]

@zane_os