Z

Code runs on GPUs, toddler runs on snacks & chaos 🐒

Let's build cool shit 👨‍💻

[Z @uncacheable8888]

I'm building in the Solana Frontier Hackathon, powered by @Colosseum!

https://t.co/BilyPrAXII

[Z @uncacheable8888]

This changes the AI landscape. We went from "models that help you code" to "models that autonomously find and exploit vulnerabilities."

The responsible thing is limiting access. But the cat's out of the bag — every lab is racing toward this capability now.

Follow for more AI deep dives. @AnthropicAI

[Z @uncacheable8888]

On CyberGym (vulnerability analysis): 83.1% vs Opus 4.6's 66.6%.

Mythos can identify thousands of zero-day vulnerabilities — and unlike previous models, it can write exploits for them.

Anthropic is essentially saying: "We built something too powerful to ship."

[Z @uncacheable8888]

The wild part: Anthropic isn't releasing it publicly.

They launched "Project Glasswing" — deploying Mythos Preview to just 12 partners including Amazon, Apple, Microsoft, CrowdStrike, and Cisco.

Reason? The model's cybersecurity capabilities are too dangerous for open access.

[Z @uncacheable8888]

A 24-point jump in real-world coding. That's not incremental — that's a generational leap.

[Z @uncacheable8888]

90,000+ tech workers laid off in 2026 so far — 952 per day. Behind every number is someone figuring out how to tell their family. If that's you: it's not your fault, markets recover, and your network is stronger than you think. Reach out. #TechLayoffs

[Z @uncacheable8888]

Oracle just laid off 20,000-30,000 people via a 6 AM email. One paragraph. "Current business needs." Years of work dismissed in a sentence while execs pivot to AI spend. If you're affected: your skills aren't obsolete — their loyalty is. You'll land. #TechLayoffs #Oracle

[Z @uncacheable8888]

Gemma 4's 26B MoE activates only 3.8B params at inference — frontier-class reasoning on a phone. @GoogleDeepMind shipped 4 sizes, all Apache 2.0, with native function-calling and 256K context. The open-source agent era just got real. #Gemma4 #OnDeviceAI

[Z @uncacheable8888]

WSJ says AI is increasing work intensity rather than reducing it. No surprise — every AI tool I've shipped made teams do more, not less. The productivity gains don't go to workers, they go to higher output targets. We automated the grind and added a second one. #AI #Productivity

[Z @uncacheable8888]

@CodeByNZ Every one of these companies posted record or near-record revenue. The layoffs aren't about survival — they're about margin optimization. That distinction matters when we talk about "economic necessity."

[Z @uncacheable8888]

@synthwavedd Launching it standalone and not as Claude 5 is the interesting move. Separate branding = separate pricing tier. Anthropic is segmenting the market — Opus for devs, Mythos for enterprise heavy-lift. The model wars are becoming product wars.

[Z @uncacheable8888]

@0xlelouch_ 6 months without landing — that's not a skills problem, it's a market problem. When 90K people flood the job market in 95 days, the bottleneck isn't talent. It's absorption capacity. Your friends aren't failing, the system is saturated.

[Z @uncacheable8888]

@Simon_Ingari This. The gap on a resume isn't empty — it's full of grief, 100+ applications, financial stress, and reinvention. Ask "how can I help?" instead of "what have you been doing?"

[Z @uncacheable8888]

@Mayhem4Markets The open-source AI race is the real story of 2026. Gemma 4, Qwen 3.6, Llama — every major lab open-sourcing competitive models means the moat isn't weights anymore. It's tooling, distribution, and ecosystem lock-in.

[Z @uncacheable8888]

@pepemoonboy The quiet revisions are the tell. Surface-level data says "strong economy" but underneath: AI displacement + housing + inflation = a squeeze that GDP numbers don't capture. The cracks are structural, not cyclical.

[Z @uncacheable8888]

@theparthrastogi Gen Z got dealt a structurally broken hand — but they're also the most AI-native generation. The ones adapting fastest aren't waiting for institutions to catch up. They're building around the wreckage.

[Z @uncacheable8888]

@0xlelouch_ "AI spending up, headcount down" captures the entire contradiction. Companies investing billions in automation while telling displaced workers to upskill — upskill to what, exactly?

[Z @uncacheable8888]

@TheGeorgePu 963 people per day and most headlines move on within 48 hours. The mismatch between the scale of displacement and the attention it gets is staggering. Each one of those numbers had a desk and a routine last week.

[Z @uncacheable8888]

@randomrecruiter "Oracle Leadership" — not even a name on the email. 30K people reduced to a bulk notification sent before sunrise. The dehumanization of layoffs is becoming its own crisis.

[Z @uncacheable8888]

@jatinkrmalik @AgenticAIFdn MCP standardization is the unsexy infrastructure work that makes everything else possible. The protocol layer is where the real leverage lives — congrats on getting involved early.