601
Agent swarms and the new model economics
We compared old and new agent swarms building SQLite from scratch and found that better coordination delivers similar quality at a fraction of the cost.
New
The latest blog posts to take off on Hacker News. 6,325 posts.
602
How we measured AI writing across arXiv, and where the measurement breaks
I scored the full text of over 12,000 arXiv papers from 2021 to 2026 across ten different fields and found that about a third of the papers currently…
603
China’s open-weights AI strategy is winning
China's open-weights AI strategy is winning: its companies are taking the lead. America's closed-first, locked-down strategy is doomed to failure - and it could take the US economy down with it.
604
Who's afraid of Chinese models?
Everyone is worried about Chinese models, but the frontier labs will be fine; we need to enable open U.S. alternatives.
605
I found a WordPress RCEs with GPT5.6 and $25
Adam Kues used a $200 GPT5.6 Sol Ultra subscription to discover a pre-authentication SQL injection and full RCE exploit chain in WordPress Core – a chain that chains a batch API desync, cache poisoning, changeset abuse, and hook injection into a complete takeover.
606
Nvidia DGX Spark as a daily driver
I bought an NVIDIA DGX Spark! It's actually quite a competent general purpose computer while being super tiny and power efficient. It basically just has an ARM CPU, a decent NVIDIA GPU (with the same number of CUDA cores as the RTX 5070), 128 GB of unified memory, and a 4 TB SSD.
607
I burned all my tokens researching how to save tokens
How I burned a full Claude limit in 30 minutes and built my own deep research pipeline instead: 3 subscriptions, shared memory, a clear role for each model. You can build the same from what you already pay for.
608
Claude Code uses Bun written in Rust now
In Rewriting Bun in Rust Jarred Sumner made the following claim: Claude Code v2.1.181 (released June 17th) and later use the Rust port of Bun. Startup got 10% faster on …
609
Ollama: All Aboard Open Models
Serving 8.9 million developers, Ollama has raised $88M from Benchmark, Theory Ventures, 8VC, Y Combinator, and many incredible angel investors.
610
AI Mania Is Eviscerating Global Decision-Making
Note: This has been cross-posted to my company's blog, in case you think there is some use in sharing with someone in a format that looks more authoritative. Link here .
611
612
The Kimi K3 Moment
I’ve been running Kimi K3 alongside Claude on my normal coding work, and for all practical purposes I can’t tell them apart. Same tasks, same quality of outp...
613
Why do AI company logos look like buttholes? (2025)
A humorous exploration of the uncanny resemblance between AI company logos and human anatomy. Discover why circular, gradient-based designs dominate the AI industry, and what this design convergence tells us about branding in tech.
614
Fable 5 vs. GPT-5.6 Sol on an NP-Hard Problem: Does /goal help?
I gave Claude Fable 5 and GPT-5.6 Sol the same unpublished NP-hard optimization problem, with and without their native /goal mode. Fable 5 is a beast; /goal is not a game changer.
615
Claude Code: Anatomy of a Misfeature
On Canada Day (July 1), 2026, Anthropic shipped a surprising "easter egg" to users of Claude Code: `2.1.198` includes an efficiency bypass which allows agents to continue on without being blocked on direction from a human.
616
Kimi K3, and what we can still learn from the pelican benchmark
Chinese AI lab Moonshot AI announced Kimi K3 this morning, describing it as their “most capable model to date, with 2.8 trillion parameters”. It’s currently available via their website and …
617
AI Meets Cryptography 2: What AI Found in OpenVM's ZkVM
We turned zkao (our AI auditor) on OpenVM, a state-of-the-art zkVM, and it found a critical soundness bug: the pairing check accepted a prover-supplied witness without proper subfield checking, which lets a malicious prover forge any pairing equality. It is fixed in OpenVM 1.6.0 and tracked as CVE-2026-46669. This is…
618
The human-in-the-loop is tired
On reward functions, dopamine, and what it actually feels like when the code starts writing itself
619
LM Studio Bionic: the AI agent for open models
The AI agent made for open models, built to get things done.
620
$100 AI Music Video: Claude Fable 5 vs. GPT-5.6 Sol
We gave Claude Fable 5 and GPT-5.6 Sol the same song, a budget, web search, and local ffmpeg, then let each autonomously direct a music video.
621
Detecting LLM-Generated Texts with “Classical” Machine Learning
.admonition { margin: .75em 0; padding: .6rem; overflow: hidden; font-size: 12px; page-break-inside: avoid; border-left: .3rem solid #42b983; border-radius: .3rem; box-shadow: 0 0.1re
622
NotebookLM is now Gemini Notebook
NotebookLM is now Gemini Notebook: the same standalone product with deeper Google integration and a secure cloud computer.
623
624
Kimi K3: Open Frontier Intelligence
Kimi K3 is the world's first open 3T-class model — frontier performance across coding, knowledge work, and reasoning, with native multimodality and 1M context.
625
The LLM Critics Are Right. I Use LLMs Anyway
I almost agree with all of the LLM critics, yet I still use LLMs a lot. I know this sounds like I am delusional, but I don't think I am alone with it.
626
Stop saying that AI is just a tool and it only matters how it is used
I’m tired of this phrase and this simple way of thinking about tools. This blog post is a wandering train of thought on the topic of what tools are and why it matters to be even slightly more mature in how we think about them.
627
LLM Networking with MikroTik
I’ve been refraining from saying ‘vibe networking’ or ‘vibkrotik’ or some other abomination, but it is true - I have been using LLMs to setup a few networks these last few months, and things have generally gone over swimmingly.
628
Why I Left Google DeepMind
I fought against Google’s Pentagon AI deal from the inside. Powerful people and institutions failed to keep their AI ethics promises under pressure.
629
Inkling: Our Open-Weights Model
Our first open-weights model: multimodal, Mixture-of-Experts, with controllable reasoning effort. Available to fine-tune on Tinker.
630
Running Gemma 4 26B at 5 tokens/sec on a 13-year-old Xeon with no GPU
A storage appliance that shipped before AVX2 existed now runs Google's Gemma 4 26B in my basement, no GPU. Here's how we got a modern MoE model onto a CPU older than the instructions its inference engine assumed were there.