The ReScript static analyzer is going incremental with the Skiplabs reactive combinators. Soon ReScript static analysis that updates in real time in the editor.

Announcing New Pareto Frontier on ARC-AGI 27.5% for just $2 333x cheaper than TRM! Beats every non-thinking LLM in existence Cost so low, its literally off the chart Vanilla transformer. No special architectures. Tiny. Trained in 2 hrs. Open source. Thread:

This is pretty crazy black magic from Cristiano Calcagno . Reanalyze runs whole program analysis to figure out things like dead code, down to the record field and variant constructor level. Incredibly powerful. In a real world project of ~1900 files and ~750k loc ReScript, doing the

Tao's compilation of info on AI's current contributions to the Erdős problems: github.com/teorth/erdospr…

This has been solved! (or so it seems?) I wanted to give a context on how amazing that is, and how many this should unlock - things that puzzled me for 10 years - but I'll just explain the solution itself, because I'm so excited about it! Turns out it is really simple, elegant,

ThePrimeagen I don’t know Greg Brockman at all but whenever a famous CTO-type person says this kind of thing, I think, OMG the entire AI sector is driven by people who don’t really understand code. It’s crazy.

When writing code was slow, teams could tolerate slower feedback loops. A 10-minute CI run felt acceptable when it took you 2 hours to write the code. When an agent writes it in 5 minutes, that same CI run is suddenly 67% of your cycle time.

Luminal fuses entire models into a single GPU kernel, automatically. Let's talk about why this matters for inference at the speed of light:

ReScript 12.1 got released: github.com/rescript-lang/…

Papers! - VeriBench: End-to-End Formal Verification Benchmark for AI Code Generation in Lean 4: openreview.net/forum?id=rWkGF… 2/n

Cristiano Calcagno n2parko Yeah, understanding how the code changed is one part. What we're trying to solve at Ketryx is even more ambitious by establishing full traceability across the req / spec architecture that goes beyond git boundaries.

Can LLMs reliably predict program termination? We evaluate frontier LLMs in the International Competition on Software Verification (SV-COMP) 2025, directly competing with state-of-the-art verification systems. AI at Meta Hebrew University Bloomberg Imperial College London UCL Jordi A.

LLMs vs the Halting Problem. (Why, what, where going.) We recently released a paper on this; link to follow. A few comments here for context. Why? With LLM "reasoning" excitement, we thought: why not try LLMs on the first ever code reasoning task, the halting problem. Turing's

Well well… ARC-AGI-2 (François Chollet’s “hardest” benchmark) is starting to smell like toast. 🍞🔥 Agentica just set a new SOTA: 85.28% with an Agentica agent (~350 lines) that writes & runs code. Best part: it’s not ARC-specialized—it's a general system that’s strong

🚨 New roles at Anthropic Zurich 🇨🇭 In addition to pre-training (where we've been hiring so far), post-training and security are joining and have open roles! It's a remarkable time in AI, the company, and on the site. job-boards.greenhouse.io/anthropic?offi…

x.com/i/article/2024…

The CSLib steering committee recently announced the official launch of CSLib — an open-source effort to formalize computer science in Lean, inspired by the impact of Mathlib in mathematics. CS researchers, practitioners, and enthusiasts are invited to get involved to support

This is awesome to watch: Agentica have solved all publically available ARC-AGI 3 tasks (mini-games)! ARC Prize It seems to work by generating bespoke program code for each puzzle. You can see it generate and progress in this video:

"Package Managers à la Carte, A Formal Model of Dependency Resolution" preprint out today: a new package calculus to describe the cambrian explosion of systems that exist today arxiv.org/pdf/2602.18602