3871
GPT-5.5
This post introduced GPT-5.5, a model built for complex tasks like coding, research, and data analysis across tools.
New
The latest blog posts to take off on Hacker News. 5,961 posts.
3872
An update on recent Claude Code quality reports
We traced recent reports of Claude Code quality issues to three separate changes. Here's what happened, what we fixed, and what we're changing.
3873
Box to save memory in Rust
dystroy - blog
3874
MeshCore development team splits over trademark dispute and AI-generated code
Migrating to the new meshcore.io site
3875
My phone replaced a brass plug
Porting a 2012 OpenCV paper to YOLOv8 and CoreML on iOS to score rifle shooting targets from a photo - replacing the brass-plug scoring gauge with on-device computer vision.
3876
If America's so rich, how'd it get so sad?
Or: How the 2020s broke our brains
3877
Math Is Hard – OpenBSD Stories
When developing software to run in a Unix environment, you will often be able to use the same system features and benefit from good developer tools, regardless of the particular platform you're working on, as most processors will provide a rich instruction set and virtual memory, among other things.
3878
3879
Writing a C Compiler, in Zig (2025)
First foray into the Zig programming language
3880
Email could have been X.400 times better
X.400 promised recallable, scheduled, self-destructing email. Why simple SMTP won instead, and what we lost and gained when it did.
3881
I am building a cloud
Today is fundraising announcement day . As is the nature of writing for a larger audience, it is a formal, safe announcement. As it should be. Writing must necessarily become impersonal at scale. But I would like to write something personal about why I am doing this. What is the goal of building exe.dev ? I am already…
3882
MacBook Neo and how the iPad should be
iPads should be radically touch only, and MacBooks should be keyboard-first
3883
Arch Linux Now Has a Bit-for-Bit Reproducible Docker Image
As a follow-up to the similar milestone reached for our WSL image a few months ago, I’m happy to share that Arch Linux now has a bit-for-bit reproducible Docker image! This bit-for-bit reproducible image is distributed under a new “repro” tag. The reason for this is due to one noticeable caveat: to ensure…
3884
OpenAI's response to the Axios developer tool compromise
OpenAI responds to the Axios supply chain attack by rotating macOS code signing certificates, updating apps, and confirming no user data was compromised.
3885
OpenAI Privacy Filter
OpenAI Privacy Filter is an open-weight model for detecting and redacting personally identifiable information (PII) in text with state-of-the-art accuracy
3886
Over-editing refers to a model modifying code beyond what is necessary
AI-assisted coding has become the norm and with tools like Cursor, GitHub Copilot, Claude Code, Codex, we are increasingly letting models touch our code. If you have used any of these tools in the past year, you have probably experienced something like this: you ask the model to fix a simple bug (perhaps a single…
3887
Workspace Agents in ChatGPT
Workspace agents in ChatGPT are Codex-powered agents that automate complex workflows, run in the cloud, and help teams scale work across tools securely.
3888
Parallel agents in Zed
From the Zed Blog: Run multiple agents at once, in the same window.
3889
We found a stable Firefox identifier linking all your private Tor identities
We discovered a privacy vulnerability in Firefox Private Browsing and Tor Browser that allows websites to fingerprint and track users across origins using IndexedDB database ordering, even after closing all private windows.
3890
Surveillance Pricing: Exploiting Information Asymmetries
For over a century, fixed prices have made markets more transparent. Surveillance pricing threatens to reverse that progress by allowing corporations to secretly tailor prices using personal data.
3891
The Joy of Folding Bikes
I was chatting to a friend about my folding bike and I had the urge to write about it - because this falls in the category of “Things I wish I’d had decades ago”. And maybe I can encourage some others to try these wonderful devices. Note: I’m 3 months into a new job so blogging has taken a back seat to drinking from a…
3892
Technical, cognitive, and intent debt
three layers of system health, AI as cognitive system 3, what's expensive if agents make coding free, and the future of source code
3893
DuckDB 1.5.2 – SQL database that runs on laptop, server, in the browser
We are releasing DuckDB version 1.5.2, a patch release with bugfixes, performance improvements, and support for the DuckLake 1.0 lakehouse format.
3894
Scoring Show HN submissions for AI design patterns
An attempt to detect AI design patterns in Show HN pages
3895
Iliad fragment found in Roman-era mummy
A papyrus fragment of Homer’s epic The Iliad has been discovered inside the wrappings of Roman-era mummy . The mummy was found in a necropolis the ancient site of Oxyrhynchus (modern-day El-Bahnasa) in Egypt’s Minya Governorate. Archaeologists were able to remove the papyrus and identify the text as the “Index of…
3896
3897
Ultraviolet corona discharges on treetops during storms
Penn State experts in meteorology and atmospheric science made their way down the nation’s eastern coast in June 2024 in search of corona discharges, a long-hypothesized atmospheric weather phenomenon where miniscule pulses of electricity dance at the tips of tree leaves, causing the canopy to glow in the ultraviolet…
3898
Qwen3.6-27B: Flagship-Level Coding in a 27B Dense Model
Qwen offers comprehensive functionality spanning chatbot, image and video understanding, image generation, document processing, web search integration, tool utilization, and artifacts.
3899
Columnar Storage Is Normalization
Something I didn't understand for a while is that the process of turning row-oriented data into column-oriented data isn't a totally bespoke, foreign concept...
3900
A 3D Body from Eight Questions – No Photo, No GPU
We trained an MLP that predicts 58 body shape parameters from 8 questionnaire inputs. The trick: a differentiable physics loss. Mass accuracy: 0.3 kg mean vs 3.9 kg for Ridge. No GPU at inference.