artursapek

It's so recognizable lol. All the staccato.

Why did he deny the plea deal? I hadn't heard that part of this story before.

why does this need to exist?

Man... PDF is such an unfortunate file format.

um my phone is not affected by this thank you very much

"Fast" was also my first impression https://x.com/tmuxvim/status/2110054212361220444

Hi HN! I've been building a word processor full time for the past ~16 months. This new NPM package provides both a CLI and TypeScript bindings to the core engine, which can be used to work with .DOCX files and other documents - with tools for redlining, commenting, converting, and reviewing diffs.

It's pretty fast, and doesn't rely on pandoc, LibreOffice, or any other external dependencies to be installed on the host system. In my tests it also lets agents get certain tasks done faster and more efficiently than they currently do.

It's essentially a headless word processor that you can build on. Would love any and all feedback.

just remember, any amount of churn rate reduction your dark patterns earn you will be paid for many times over in the afterlife

New York Times in shambles over this one

Why do Rust developers have to say shit like this

should I open it up? :)

What’s your problem with spacex? Do you have Elon derangement syndrome?

In the niche benchmark I run it’s on par with 6-Sol, not close to Astra.

6-Luna dropped it to 1c since you wrote this! :D

Cracker news upvotes the stupidest shit to the front page

Shameless plug of my app! https://revise.io/editor

It's a word processor that lets you selectively accept AI suggestions, or just talk to the agent in margin comments. I've been working on this full time for over a year now.

I only use OpenRouter to benchmark models. For my production app, I have direct integrations with OpenAI, Anthropic, Google, xAI. Even just input caching is enough of a reason to do that, assuming you are trying to build the fastest and most cost-efficient product possible. I don't understand how people supposedly run production apps using OpenRouter unless those two variables somehow don't matter to them.

Yes, it uses block IDs - documents are made of blocks like paragraphs and lists. Formatting survives because the agent works in HTML.

Hi HN. I've been bootstrapping this project full-time for the last 12 months. Would love to get some feedback on the MCP integration! I think it's some of the best UX available for working on documents, with AI.

These prices are not real. They already said so.

It’s an attempt to build a locomotion model. It uses a physics engine (Avian in Bevy) to animate a walking biped from first principles.

An earlier version can be seen here https://x.com/tmuxvim/status/2081822899149963422

WASD or just touch on mobile

It's definitely not cheaper than Sonnet on my benchmark, but it's cheaper than Fable and outperforms it. Which is big IMO. https://revise.io/errata-bench

HN users are world champions are trivializing difficult things with snarky comments

Yep, I've been taking glycine and magnesium for years. I am not as consistent as I should be but it makes a big difference when I use them.

I run a proofreading benchmark that tests how well models can find and fix errors in English text. They get several passes in a simple agent loop. Sonnet 5 is definitely better than Sonnet 4.6, but inferior on both quality and cost to GLM 5.1, GLM 5.2, Gemini 3.1 Flash, and Gemini 3.1 Pro. https://revise.io/errata-bench

haha yeah I've bet the last 12 months of my career on a .io

The .ai TLD is some tiny island with a few thousand people

Trivial to simulate basic keystrokes. But I don't think it's trivial to simulate the natural process of drafting something. There's no concrete heuristic or algorithm (yet) for judging these types of replays, but I'd be impressed if someone can actually make a program that reliably simulates a natural keystroke-by-keystroke thought process which appears human when replayed in this way.

They claim extreme performance on ExploitBench, which Mythos was touted as being incredible at. https://x.com/OpenAI/status/2070555278576439306