The Dark Night of Mathematics

Engineering management after the cost of code collapsed

Claude Opus 5

Claude Opus 5

Hetzner is working on LLM Inference

Nothing works and everyone is euphoric

Kimi K3 exploited the latest Redis server

I Think You Might Be Fooling Yourself with AI

Petals: Run LLMs at home, BitTorrent-style

OpenAI’s accidental attack against Hugging Face is science fiction that happened

Protecting our FLOSS commons from LLMs

GigaToken: ~1000x faster Language model tokenization

Are AI Labs Pelicanmaxxing?

Can a MUD evaluate LLMs? A $99 proof of concept

So Reddit has decided that plain HTML is unsafe

Codeberg bans vibe coded projects

Kimi K3: second only to Fable 5 on AA-Briefcase

Gemini last models: temperature, top_p, and top_k are deprecated and ignored

Advertise in ChatGPT

Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber

Gemini 3.6 Flash

Controlling Reasoning Effort in LLMs

Qwen3.8 is launching and going open-weight soon

What's the deal with all the random weekly quota resets for agents lately?

GPT-5.6 used a prompt to close a 30-year gap in convex optimization

Kimi K3, and what we can still learn from the pelican benchmark

The human-in-the-loop is tired

German AI consortium releases Soofi S, an open 30B model that tops benchmarks

Detecting LLM-Generated Texts with “Classical” Machine Learning

Generative AI Is an Engineering Disaster

The LLM Critics Are Right. I Use LLMs Anyway

LLM Networking with MikroTik

Running Gemma 4 26B at 5 tokens/sec on a 13-year-old Xeon with no GPU

Towards a Harness That Can Do Anything

DSLs Enable Reliable Use of LLMs

Solving 20 Erdős Problems with 20 Codex Accounts Running in Parallel

Bonsai 27B: A 27B-Class model that runs on a phone

Guardian Angels: LLM Personalization for Productivity and Security

Coding agents think ahead of time

Punch yourself in the face with reality

Control the Ideas, Not the Code

I love LLMs, I hate hype

Mechanistic interpretability researchers applying causality theory to LLMs

Fixed three bugs that made Qwen3.5-122B a daily driver on Mac Studio

Mesh LLM: distributed AI computing on iroh

Stop Telling Me to Ask an LLM

Please don't discontinue Gemini 2.5 Flash

Write code like a human will maintain it

Unified Memory, Explained: Why Mini PCs Can Run 70B Models a Big GPU Can't

I think I have LLM burnout