Black_WillyWonka

Moi

[Black_WillyWonka @blackwillywonka]

When your custom Claude setup already detects desk manipulation on Hyperliquid in the first few hours…
👀
#Claude #Anthropic #Hyperliquid #Crypto #AI #DeFi #OnChain

[Black_WillyWonka @blackwillywonka]

Quand ton Archi AI te met en évidence, dès les premières heures d’observation sur Hyperliquid, des Desks en flag de manipulation de cours…
👀
#Claude #Anthropic #Hyperliquid #Crypto #AI #DeFi #OnChain

[Black_WillyWonka @blackwillywonka]

@HerveH67917 @stonech19 @Romain_Molina

[Black_WillyWonka @blackwillywonka]

@Aurelien_Le_Coq c'est dans quel but d'envoyer des DMCA à des youtubeurs?
Je ne connaissais pas mais quand on m'a envoyé que vous attaquez Réaction Révolution un petit créateur de contenu sur YT pour droits d'auteurs (d'une émission E1)... C'est soit une erreur soit une erreur...

[Black_WillyWonka @blackwillywonka]

@doseche @FraPiquemal @J_Bardella Depuis que les notes de rac de Bardella ont leak il faut reconnaître qu'il ne prend pas beaucoup de risques avec cette déclaration 🤣🤣🤣

[Black_WillyWonka @blackwillywonka]

I can't wait to see this in recruitment sessions 🤣🤣🤣

[Black_WillyWonka @blackwillywonka]

Welcome to NICE.
Same agents.
New smile.
Same nightmares.

Stephen Miller approves this message.

#ICEisNICE #Trump #BorderPatrol

[Black_WillyWonka @blackwillywonka]

@mwfowlie @subz3r0___ @AnthropicAI I tweeted about it a few days ago it's too frickin much.

[Black_WillyWonka @blackwillywonka]

24 hours after my weekly reset on Claude Max x20…I’m running only TWO Claude Code CLI instances in my IDE purely as xhigh orchestrator for complex multi-agent workflows with 1M context.Already:
→ 100% of weekly quota completely torched
→ €300.15 spent on extra usage
1) This token burn is completely insane and unsustainable. At this rate, it’s already cheaper to buy 5 maxed Mac Studios (512 GB unified memory) right now than to keep paying Anthropic with these constant extra charges every 15 minutes.Open-source is closing the gap fast.

[Black_WillyWonka @blackwillywonka]

Claude Opus 4.7 on Max x20 is straight-up painful.
I only run TWO instances of Claude Code directly in my IDE — purely as orchestrator in xhigh for complex multi-agent workflows.
Less than 24h into the week → already 95% of my quota burned.
Even when it “just” maintains context and calls other agents.Open-source models are now close enough.
At this rate, it’ll soon be cheaper to buy 5 maxed out Mac Studios (512 GB unified memory) and run everything locally than to keep paying Anthropic + watching extra usage charges hit my card every 15 minutes.This token economy is broken.

Claude getting too expensive? What are you doing?

Who else is eyeing local?
What’s your current local setup right now? #ClaudeAI #Anthropic #LocalLLM #MacStudio #ClaudeCode #AI

[Black_WillyWonka @blackwillywonka]

@auroreberge @LEXPRESS https://t.co/3MpNLZrNcG

[Black_WillyWonka @blackwillywonka]

@kenaz_ai @bcherny Congrats you're the second occurrence of the outage I could find on X

[Black_WillyWonka @blackwillywonka]

@BasedInfra @nizzyabi Congrats you're the first occurrence of the outage I could find on X

[Black_WillyWonka @blackwillywonka]

@au1848 @hypnotiik92 @knafo_sarah Et pourtant sont commentaire reste toujours aussi pertinent...

[Black_WillyWonka @blackwillywonka]

@512banque @D_nAAkam0t0 BMAD m'a changé la vie
J'ai perso un peu le workflow mais après un peut moins d'une semaine de setup et de cadrage, je part de chez moi et je reviens yh après et mes agents bossent toujours sans avoir dévié ou très peu.
Quelques review et c'est reparti pour 6h de plus.!

[Black_WillyWonka @blackwillywonka]

@Sully_T I don't give my workflow for free but I can tell that when I need verbal creativity and human understanding I use GPT5.2, for web design Gemini pro and flash in thinking mode are m'y go to and coding is uncontestable it's Claude 4.5 Opus despite the fact that it's pricey