Supply chain sustainability practitioner. 17 yrs, 4,700+ suppliers transformed. Writing Wiweck on Substack, cutting through ESG noise for business leader
Is anyone observing, #Claude Opus has begun consuming credits far faster than before.This sudden, accelerated credit usage is abnormal and suggests a potential issue with Claude Opus. @AnthropicAI@claudeai
Where was this fire, this urgency, this collective conscience when education was being stolen from the poor, from Dalits, from minorities, from the marginalized? #education
https://t.co/gTNjOOu1Bh
Where was this concern, this urgency, this solidarity when the very right to education was being quietly stolen from the poor, the Dalits, the minorities, and the marginalized?
https://t.co/23QMRZYdK9
Tried Manus for deep research. Big mistake. A single prompt instantly vaporized 4,000 credits. It’s not a tool; it’s a predatory cash grab. Completely unsustainable for complex work. @ManusAI
@kingkyle9@ManusAI US AI models are becoming economically unviable for complex, real-world tasks. The token economics simply don't support sustained real work workflows. I’m thinking to move my heavy lifting to Chinese open-source models.
@claudeai@AnthropicAI Fable oversensitivity makes no sense, if a non-coder can't use it for basic idea brainstorming or research, what's the actual point of the product?
Societies under material stress tend to produce both greater religiosity (as a coping mechanism) and greater social cruelty (as a resource competition response).
@GarvitSura5238 whereas the regular token allotment easily supports 10–12 prompts. This suggests a potential flaw in how Claude measures extra token consumption.
$35 in Claude extra usage credits disappeared after two detailed prompts. This is not a quirk; it reflects how extra usage actually works. Unlike your plan allowance, extra usage charges at full API rates, where long outputs and deep context windows compound cost fast.
@GarvitSura5238 Agree. My work focuses on policy rather than technical tasks, which require extensive reading and contextual analysis. This consumes tokens quickly, especially under the extra token allocation. On a regular subscription, token usage feels much more reasonable.
@GarvitSura5238 It seems claude uses a different calculation method for regular versus extra tokens — regular tokens deplete sensibly, while extra tokens get exhausted disproportionately fast. For instance, two prompts in Sonnet can burn all extra tokens,
@GarvitSura5238 I don't use any technical tools. I use Sonnet exclusively for regular tasks. This is not the first time it has happened. Last month, the same issue occurred. In just one prompt, the entire weekly usage burned.
A single comprehensive exchange can consume hundreds of thousands of tokens once conversation history is factored in. Anthropic does not make this pricing asymmetry obvious at the point of purchase. @AnthropicAI@claudeai
When 20% of world oil and gas passes through a 21-mile strait and that strait closes, every import-dependent economy asks the same question:
What would it cost to never be this exposed again?
India. Japan. South Korea. Most of Europe. All asking it simultaneously.
The petrodollar is the quiet casualty.
China is settling Iranian oil purchases in yuan, transaction by transaction, building structural alternatives to dollar energy trade.
Every month this conflict continues, that shift becomes more permanent.
No climate summit delivered the argument for renewables that this war just did in five weeks.
Energy independence is now a survival argument, not a sustainability one.
And that changes everything about who is listening. 🧵