Sunny

Interesting:

🚨 THE DOMINOES ARE FALLING. IN DAYS, NOT WEEKS. Here is the global oil crisis — country by country — as of today: 🇱🇰 Sri Lanka — RATIONING. 4-day work week. Schools closed. 🇵🇰 Pakistan — CRISIS. Overnight price surge. Long queues. 4-day work week. 🇮🇳 India — 9 DAYS of reserves left. Emergency suppliers being hunted. 🇰🇷 South Korea — 50 DAYS left. Clock ticking. 🇯🇵 Japan — LIED. Said 254 days. Actual usable reserves: 95. 🇬🇧 UK — Shell CEO warning: SHORTAGE STARTS IN APRIL. 🇩🇪 Germany — Gas up 30%. EU emergency plan launched. 🇫🇷 France — Paying 30% more at pumps than 8 weeks ago. 🇿🇦 South Africa — Government says "stable." Citizens photograph empty pumps. 🇹🇷 Turkey — Stocks crashed. Inflation exploding. Currency under pressure. 🇧🇷 Brazil — Watching nervously. Own oil helps but supply chains hurting. 🇦🇺 Australia — Tanker delays. Import dependent. Hoping IEA coordination holds. 🇺🇸 USA — Gas taxes suspended in states. SPR drawn down. Iran sanctions quietly paused. 🇨🇳 China — 1.4 BILLION barrels stockpiled. Banned exports. Still getting Iranian oil. This is DAY 26 of the Hormuz blockade. 20% of global oil — GONE. 8 million barrels per day — GONE. OPEC's response: +206,000 barrels. That is 2% of the hole. Here is what nobody is telling you: They're showing you "reserves are sufficient." They're NOT showing you Japan overstated reserves by 3x, South Africa has empty pumps, and India has 9 days. The real energy crisis hasn't even started yet. RT so this doesn't get buried by the algorithm.

he is speed talking now 🤣

A young girl practicing martial arts while waiting for the bus. Sometimes, you need less than you think to stay active.

😵‍💫

New on the Engineering Blog: How we designed Claude Code auto mode. Many Claude Code users let Claude work without permission prompts. Auto mode is a safer middle ground: we built and tested classifiers that make approval decisions instead. Read more: anthropic.com/engineering/cl…

中国・北京🇨🇳から流出した映像。

@levelsio Cool

Okay let's see who can reply to this

@bridgemindai cache hit not working

I just hit my Claude Code 5 hour limit insanely fast. What is happening?

@sudoingX Dont think its better than 27b

it's fast and smart. let me dive deeper and write for you all.

Woooo

Hey @grok bring this picture back into Life

Casual Bugatti Chiron sighting on Cambodian TikTok. Starting price around $3 million. Cambodia’s GDP per capita is about $2,800. Via @monyy88887 on TikTok

“Are you free after work?” me after work:

These geckos are so cute! 😍 😍

@sudoingX have u tried huggingface.co/Jackrong/Qwen3… ?

first impressions of qwen 3.5 27B dense on a single RTX 3090. 35 tok/s. from 4K all the way to 300K+ context. no speed drop. hermes 4.3 started at 35 and degraded to 15 as context filled. qwen dense holds. MoE held 112 flat. 3x faster but only 3B of 35B active per token. architecture tradeoff. Q4_K_M on 16.7GB. native context 262K. pushed past training limit to 376K before VRAM ceiling on 24GB. tried q8 KV cache at 262K, speed collapsed to 11 tok/s. q4_0 KV is the sweet spot. flash attention mandatory. built in reasoning mode. the model thinks step by step before it answers. full chain of thought surviving Q4 quant. 1,799+ token thinking chains with self correction loops. on a single consumer GPU. gave it one prompt: "build a realtime particle galaxy simulation in one HTML file." 3,340 tokens. 95 seconds. one shot. ran on first load. full reasoning and coding in the video below. optimal config if you want to skip the hours of testing: llama-server -ngl 99 -c 262144 -fa on --cache-type-k q4_0 --cache-type-v q4_0 this is just the warmup. octopus invaders is next: 10 files, 3,400+ lines, zero steering. the prompt hermes quit at 22%. already more impressed than expected. full results coming soon.

Animal voice prank compilation 😂😂

We’re launching with two new posts. Can AI do theoretical physics? Harvard physicist Matthew Schwartz led Claude Opus 4.5 through a graduate-level calculation. AI can’t yet do original work autonomously, but it can vastly accelerate it. Read more: anthropic.com/research/vibe-…

Software horror: litellm PyPI supply chain attack. Simple `pip install litellm` was enough to exfiltrate SSH keys, AWS/GCP/Azure creds, Kubernetes configs, git credentials, env vars (all your API keys), shell history, crypto wallets, SSL private keys, CI/CD secrets, database passwords. LiteLLM itself has 97 million downloads per month which is already terrible, but much worse, the contagion spreads to any project that depends on litellm. For example, if you did `pip install dspy` (which depended on litellm>=1.64.0), you'd also be pwnd. Same for any other large project that depended on litellm. Afaict the poisoned version was up for only less than ~1 hour. The attack had a bug which led to its discovery - Callum McMahon was using an MCP plugin inside Cursor that pulled in litellm as a transitive dependency. When litellm 1.82.8 installed, their machine ran out of RAM and crashed. So if the attacker didn't vibe code this attack it could have been undetected for many days or weeks. Supply chain attacks like this are basically the scariest thing imaginable in modern software. Every time you install any depedency you could be pulling in a poisoned package anywhere deep inside its entire depedency tree. This is especially risky with large projects that might have lots and lots of dependencies. The credentials that do get stolen in each attack can then be used to take over more accounts and compromise more packages. Classical software engineering would have you believe that dependencies are good (we're building pyramids from bricks), but imo this has to be re-evaluated, and it's why I've been so growingly averse to them, preferring to use LLMs to "yoink" functionality when it's simple enough and possible.

LiteLLM HAS BEEN COMPROMISED, DO NOT UPDATE. We just discovered that LiteLLM pypi release 1.82.8. It has been compromised, it contains litellm_init.pth with base64 encoded instructions to send all the credentials it can find to remote server + self-replicate. link below