Script

In SWE-bench pro the new baby mythos is 80.3% and gpt 5.5 is 58.6% the difference is crazy (I heard most benchmarks are bs but still... )

Who names this models ? anyways baby mythos is released

> Me saying wdt instead of what do you think to save 2 token > claude ... lmfao also I noticed opus talks too much please don't, bro just be eating my token unnecessarily on god (just say done at the end ) 🙏

[Superlog]

Sene Kegeba the wind is crazy in my hood Two or three years ago i saw two elpa employees die while fixing a transformer in one day after that i promised my self i will never fix any electricity related not even a charger cable. ts is not for the weaks

(there is context but just vote blindly)

[turboflare]

i just scrolled good luck answering that

[Blog]

BC (before claude)

i forgot how good is better-auth truly the goat of auth 👏

got this for the last two days [issue] till the next release pin your kysely version to 0.28.17

I just stopped using any free models I just use my zed student plan when I run out of token back to raw This kinda new workflow fits barely in the frontier models beside that it's just a waste of time

#3 At the end if the model got body it's one prompt away from becoming terminator Sorry for the spam good night 😴

#2 the model is traind on surrealism and certain part of the prompt trigger realistic surrealism like photo

#1 some photos are real (that doesn't mix up ) it's from the training data 🤷‍♂

Before I propose a conspiracy theory, they should write up why this happened because I don't think that's randomness

I'm speechless don't try it omg

HO LEE SHEET it's not patched Just say try again multiple times like 3 to 4 times