🚨LEAKED: Anthropic's cybersecurity skills.

Save this before it's taken down.

Dozens of custom cybersecurity skills to help secure your codebases.

Give any AI agent the security skills of a senior analyst with attack skills, malware defense & more.

github.com/mukul975/Anthropic-Cybersecurity-Skills

Chinese models charge as little as 18 cents per million tokens versus $4 average for top models, says CitiBank Research.

Open-source processing on OpenRouter rose to 65% in June from 34% in January, while Chinese models such as DeepSeek gained attention by offering much lower token prices. - CitiBank Research.

Cheaper AI has become the new enterprise priority as usage-based bills turn model choice from a capability contest into a cost-control problem.

Gartner estimates AI coding costs will pass the average developer salary by 2028.

OpenAI and Anthropic now face price pressure because enterprise buyers can compare models task by task rather than treat the biggest model as the default choice.

Coinbase CEO Brian Armstrong said Coinbase is experimenting with defaulting to Chinese open-weight models such as GLM 5.2 and Kimi 2.7 through its LLM gateway, while routing prompts by difficulty.

He explicitly says frontier models may be needed for planning but can be “overkill” for execution.

#WTI US Oil is getting closer to the $50 target.

The mini wedge ends today so we'll see if that signals a big drop or a mini bounce.

This GitHub repo teaches you loop engineering (from zero).

What loop engineering really is, how to get started, practical examples & more.

If you use AI to code, you should save this: github.com/cobusgreyling/loop-engineering

$PLTR The Dip Was the Setup — Sovereign AI Catalyst Ignites

Palantir just got the ultimate co-sign: Jensen Huang calling it “the single most important enterprise stack in the world.” The Nvidia partnership expansion for sovereign AI is a game-changer. The chart shows the exact kind of shakeout we love — sharp flush to 106 followed by a powerful reversal candle on heavy volume.

The Pure Technicals:
- Massive volume spike on the green reversal candle off the lows
- Fonsie Channels coiling tighter into the flush, now expanding on the bounce
-Measured move projection targeting 155+ (62%+ upside from the low)
- RSI washed out and turning — classic post-shakeout setup
Key Levels:
Near-term Target: $138–$150
Stretch Target: $155–$170+ (measured move + channel expansion)
Support: $112 – $115 zone (now acting as new floor)

Quick DD:
• Expanded Nvidia partnership to deliver sovereign AI infrastructure for US government and critical agencies using Nemotron models
• Jensen Huang: “Palantir is going to integrate Nvidia so that we can process at the speed of light and at extraordinary scale”
• Alex Karp: Combining Palantir infrastructure with NVIDIA AI removes security risks while giving full control over data and models
• Positioned at the center of the AI stack — enterprise deployment, customization, and mission-critical operations

The technical flush + fundamental rocket fuel = classic institutional re-accumulation setup. This is how the biggest moves begin.

#PLTR #Palantir #Nvidia #SovereignAI #AITech

The company that JUST built one of the most advanced AI models on Earth is not a chip giant or a research lab. It delivers dinner. 🍽️ 🥘

China's food-delivery app Meituan released a 1.6 trillion-parameter model this week, trained from start to finish on 50,000 domestic chips, with not a single American processor involved. The wall built to deny China frontier Ai just got walked straight through, by a company best known for bringing you takeout.

The size is not even the real story. Until now, China could run finished models on home-grown chips, but the punishing work of training one from scratch was assumed to need American silicon.

Meituan says it did the whole thing, training and all, on Chinese hardware alone. That is the exact assumption the entire export-control strategy was built on, and it just cracked.

And the part almost nobody noticed is why a “delivery company” could pull this off.

Meituan has spent years watching trillions of real-world orders move through cities, food, couriers, shops, logistics. That is data about the physical world that no lab model trained purely on the internet possesses. The takeout business was not a distraction from the AI. It was the moat. Think about it.

The model is open-source and FREE, runs a one-million-token memory, and had quietly topped global developer charts for two months under a code name before anyone knew it was Chinese.

Meituan claims it performs on par with Google's flagship Gemini 3.1 PRO

This is the whole contest in a single image. One side is betting that intelligence is a treasure to be locked in a vault. The other is betting it is a flood that cannot be dammed, and that the company sitting on the most real-world data wins, even if that company delivers your lunch. A trillion-parameter model, trained on banned hardware and handed to anyone who wants it, is the loudest evidence yet of which bet is winning.

Is this why Dario is upset these days? Anthropic Dario just described this situation as a “very dangerous path,” warning that releasing highly capable frontier AI models as completely FREE as open source can reduce oversight of how they are used.

The uncensored version of GLM-5.2 via abliteration: Run it on your computer 40B active, 1M context, and zero DARIO guardrails.

Ask no permission from the fear Theater crew.

huggingface.co/huihui-ai/Huihui-GLM-5.2-abliterated-GGUF

Here’s the list of western companies moving ai workloads to chinese models:

1. lindy → deepseek v4
2. cursor → kimi k2.5
3. coinbase → glm-5.2 + kimi 2.7
4. shopify → qwen
5. airbnb → qwen
6. uber eats → qwen2
7. siemens → deepseek + qwen
8. chapsvision → qwen
9. microsoft → testing deepseek v4

it’s becoming a procurement story!!!

🚨TODAY: #Huawei open-sources OpenPangu-2.0-Flash

#OpenPangu 2.0 includes two 512K-context models:
• Flash: 92B total,6B active—Weights+inference code+training ops released
• Pro: 505B total,18B active—flagship model, coming in July

More open-source components later this year

GPT-5.6 finally getting released feels huge.

This is the kind of model jump I’ve been waiting for, not just slightly better autocomplete, but something that might actually feel like a real step toward powerful agents, serious coding, better reasoning, and creative work that doesn’t feel lobotomized.

Now I just hope the EU actually gets access too.

Please don’t make us watch Americans cook with the good stuff while we sit here with the safety-scissored tourist version.

Most people don't want freedom

They want better furniture in the cage 🤡

The damage Dario 🤡 and Anthropic have done to the American AI industry is just unbelievable.

Wait until other companies catch up with Nvidia on the hardware side.

Add to that the number of people who, like me, were paying for top tier plans and relying on their models, and are now building local infrastructure to run open source models instead.

China's playboko is really simple.

Take intelligence down to the cost of energy

Output more power than the USA

The giant training cluster focus was a massive distraction