Its not like that shit is anything revolutionary, its just a cli tool with fancy formatting

>>108491845 (OP)
who in the hell cares about this.
the juice is in the models, not some shitty ui front end (now commodity).

>>108491845 (OP)
Ok and? It's a basic CLI app.

>>108491845 (OP)
Okay, it's fukken saved. What the fuck did I just download?

Can I swap the API calls to use local LLM?

>>108491845 (OP)
They are recording your insults so when singularity is here you'll be one of the first ones taken out
https://github.com/chatgptprojects/claude-code/blob/642c7f944bbe5f7e57c05d756ab7fa7c9c5035cc/src/utils/userPromptKeywords.ts#L8

isn't opencode or whatever the name was the same shit? The model weights would be far more interesting but thats like several tb so very hard to leak.

>>108491845 (OP)
Should've told it to make no mistake.

>>108491967
>Can I swap the API calls to use local LLM?
bruh you don't need the source code for that, just set the ANTHROPIC_BASE_URL environment variable

>>108491967
Just use OpenCode...

>>108491983
>several tb so very hard to leak
i'm surprised none of the big labs models have leaked yet considering the weights are duplicated across millions of data center computers
i suppose people fear the inevitable lawsuit and jail time

>>108491845 (OP)
>no models
So it's worthless?

>>108491987
How would I know? I don't use that, just hoarding it for a rainy day, yknow.

>>108492006
Yeah, it's almost like data center security is taken seriously or something.

>>108492026
physical security yes.. but network security i wouldn't be so sure

Guys someone leaked codex as well
https://github.com/openai/codex

This is big

>>108492006
Anthropic being smaller company yet beating the competition kinda hints you they might be good with security too.
>>108492026
No, it's actually a joke. Often random people would end up there and such.

>>108492007
>>no models
>So it's worthless?
it would be worthless even if models leaked. which one of us has a DGX cluster worth 500k at home to run the models?

>>108492062
Don't be an idiot. Chinese and Arabs would make you very rich if that model came up to them in a dream, you know what I mean? So that they could "develop" their own model and deploy on their hardware and made it available for pennies.

I personally would not do that only because I hate Russians too much to give them access to something like that for free. I'd rather pay Anthropic $20 monthly till the day I die.

>>108492077
the weights would be useless to them for developing their own models, in that case you'd need to leak the whole training pipeline

>>108492099
They could have original one, which is very good as it is. And then distilled ones, fine-tuned ones etc.
If you have current Anthropic top model, you would not need anything else in the nearest future, in maybe 3 years.
You're an Arab that invested in ASICs for LLM inference, you got your hands on these models -> make crazy money for years. So much so, that you would actually turn this shit profitable. For your company only ofc, there's no saving AI bubble in general.

>>108492077
>Chinese
Have their own models and research teams responsible for multiple AI breakthroughs.
>Arabs
Are currently running around with their asses on fire looking for air defense to protect their oil and desalination infrastructure from the lawnmower drones invited by the Americans.

File: 1753409699415507.mp4 (744 KB, 720x720)

744 KB MP4

>Trump gives up on Iran
>Claude code leaks

03/31/2026

TACO TUESDAY

>>108492006
And what would you do with such a model, none can even run it without a datacenter i guess

>>108491845 (OP)
Hmmm

>>108492040
You can laugh all you want. I just downloaded that .zip as well.
Noticed the amount of commits afterwards.

the vibes were too strong