Meziane I. Mahieddine

i write code sometimes, i procrastinate most of the time.

[Meziane I. Mahieddine @maa7io]

I'm a noob in Mobile app developement, i have never built apps for mobile iOS/Android.

Today I'm going to be starting a dev journey to learn and build my very first android app and will be documenting my journey here and updating this thread daily to track my progess.

[Meziane I. Mahieddine @maa7io]

And for ASI, the model will no longer rely on TEXT for it's reasoning but it will use pure math for everthing and only it can understand it's own reasoning and output.
Because text is for humans and robots will find a much better way to communicate both internally and externally.

[Meziane I. Mahieddine @maa7io]

Fable 5.2 and Mythos 5.2 are both impressive models and very capable but i'm here for the folks that overhype every newely release model from @AnthropicAI and calling it AGI or some even called it ASI.

Guys chill...

Continue reading 👇👇

[Claude @claudeai]

During pre-release testing, Claude Fable 5.2 and Claude Mythos 5.2 worked on ambitious projects in many different fields. Here are a few.

From a one-paragraph prompt, Fable 5.2 designed three working electronic devices, including this home voice assistant hub.

[Meziane I. Mahieddine @maa7io]

It has to have multiple ways to interact with the real world, not only through text, but through "internal thoughts" just like our brains.

The current reasoning has to be improved alot so that it's faster, more accurate and concise so it can make decisions on a whim.

[Meziane I. Mahieddine @maa7io]

POV: You type too fast after three coffees 💀
@npmjs please protect us from ourselves

#ProgrammerHumor #CodingMemes #DevLife

[Meziane I. Mahieddine @maa7io]

@0xMovez Can you share how long it took to create and how many tokens/dollars it costed you to create this animation?

[Meziane I. Mahieddine @maa7io]

@JamesMalsawm CENSORSHIP and paying alot for the 'SAFETY'

It's not like GPT-6.1 is open but you get way less refusals than in anything made by Claude/Anthropic.

[Meziane I. Mahieddine @maa7io]

@pseudosaxon @Solvairo @chasmmmmmmmmmmm Then why does claude when commiting to GIT adds it self as a contributor?

Who actually owns the output and if anything wrong happens who is liable?

[Meziane I. Mahieddine @maa7io]

@MyMoonEnt @sdmat123 @JesusIsGod @chasmmmmmmmmmmm Who do you think is paying for that extra Haiku Inference used for safety?

I don't think Anthropic is...

[Meziane I. Mahieddine @maa7io]

This is after a 2 hours run on a ~100K LOD code base.

notice how the messages are holding lots of context which in turn ends up buring through tokens very fast.

if we tell #Claude to send as little messages as possible to conserve context, will that lower the output quality?

[Meziane I. Mahieddine @maa7io]

@CtrlAltDwayne Not just opus, the overall 5-Hour and weekly allowance shrunk and they are giving us way less than what we used to get.

Notice that even Sonnet 5.5 now is buring through tokens like there is no tomorrow.

[Meziane I. Mahieddine @maa7io]

@pcshipp The Pro plan is literally useless.
The 5-hour session barely gets you 20 Sonnet 5.5 Prompts, not even in Code but in chat.

Claude has been nerfed since that $80B loss anounced by @AnthropicAI in the first half of 2026.

One Opus 5.5 prompt burns through 30% of 5-hour session

[Meziane I. Mahieddine @maa7io]

@WukongNumber1 @AnthropicAI @claudeai Yeah!
Previously i used to make micro adjustments to the code base using Sonnet and it would last long enough, with around 10 to 20 Opus long runs for big features, refactoring or general code base audit and unit tests.

Now i can barely get through the week with sonnet only.

[Meziane I. Mahieddine @maa7io]

Claude Code on a Pro Plan is completely useless...

The weekly allowance barely gets you through 4 days of real work.
@AnthropicAI @claudeai

[Meziane I. Mahieddine @maa7io]

Youtube in incognite mode will give you better suggestions, espeacially in music.