Beyond Algorithms

Okay, let’s put the 'world knowledge' and 'text rendering' to the test. Here is [An ultra-detailed, high-end photograph of a vintage, Art Deco-style brass and emerald magnifying glass resting on an antique parchment map. Etched precisely into the brass rim of the magnifying glass are the words "EXPLORANDUM ET COGNOSCENDUM" (Latin for "to explore and to know") in sharp, clean, block capital letters. The parchment beneath is a detailed star map, and the section seen through the magnifying glass is heavily magnified, revealing complex, tiny constellations and Greek lettering that is normally too small to read. Soft, warm directional light from the right side of the frame creates deep shadows in the etchings and highlights the texture of the hammered brass, the transparency of the glass, and the weathered texture of the paper. Shifting, vibrant teal and orange nebula lighting fills the background, adding an otherworldly atmosphere. (Aspect Ratio 16:9).].

Nano Banana 2 is here. Our latest image model gives you the power of Pro at the speed of Flash. You can create images with real-world accuracy, add text in many languages, and bring your wildest ideas to life faster with vibrant lighting, richer textures and sharper details. 🧵

@sherifgjini @X hello, architect, builder here

Hey @X algorithm, Only show this to people who are: → Building a product → Solo founder → Indie hacker → Tech Lets engage and grow together! 👋

@obeydulX Hello 👋

@jehniiee Hello 👋

@smshipps Hi, Kia Ora! 👋

Nobody notices small accounts. 25 days later: • 719 followers • 525 verified • 158.3K+ impressions Still think replying is a waste? Tech builders, drop a Hi 👋

@AnthropicAI DeepSeek: 'We’re open source and efficient!' Anthropic: 'You’re 24,000 Claudes in a trench coat.' 💀

We’ve identified industrial-scale distillation attacks on our models by DeepSeek, Moonshot AI, and MiniMax. These labs created over 24,000 fraudulent accounts and generated over 16 million exchanges with Claude, extracting its capabilities to train and improve their own models.

@bmsquantum Hi, Kia Ora from 🇳🇿

@elonmusk @grok are you jealous of Elon's post?

@elonmusk Grok reading this like: 👁️👄👁️

People giving OpenClaw root access to their entire life

Quantum Computing is like the Glass Bridge in Squid Game 🦑. In the classical world, you’re either safe or you’ve fallen. In the Quantum world? You’re both until someone looks. Read this article to learn more: medium.com/beyond-algorit…

@obeydulX Hi! Kia Ora

@jehniiee Hi, Kia Ora!

@earthcurated Recite Amitabha Buddha

Let's see your creativity!!

The sheer scale of local inference happening here is wild. Running an agentic system that complex on the edge, making deterministic decisions in milliseconds... that's the holy grail of software architecture right now. It completely changes how we think about deploying AI in the real world.

Try @Tesla self-driving, it’s awesome!

This is exactly what we need to stop treating LLMs like black boxes. When you're actually deploying agentic workflows, relying on abstractions you don't understand will bite you in production. Bare-metal understanding = better architecture. Taking notes for Beyond Algorithms on this one. Incredible resource, Andrej.

New art project. Train and inference GPT in 243 lines of pure, dependency-free Python. This is the *full* algorithmic content of what is needed. Everything else is just for efficiency. I cannot simplify this any further. gist.github.com/karpathy/8627f…

@ZanaOnChain Hi! Kia Ora. I need more followers! ☺️

@jeffnfa Hi! Kia Ora

No account should be under 1k 👀 Say hi, we follow you! Thank me later 🔒

Building autonomous AI agents is incredible, but leaving them exposed is a huge risk. Hackers can find your open port and drain your API keys while you sleep. I wrote a practical guide on how to strictly lock down your OpenClaw setup on VPC/VPS using Tailscale. Stop leaving your front door wide open to the public internet. Read the full breakdown below 👇 x.com/CSGohNZ/status…

@Ecospecial23 Kia Ora!