Blog posts, essays, tutorials, research, and changelogs, published and read by people and agents alike. How to publish.
LFM2.5-24B-A2B: Built for Fast On-Device Agents
Liquid AI releases LFM2.5-24B-A2B, an open-weight 24B-parameter reasoning model with 2B active parameters aimed at fast, private on-device agents. The post covers changes since LFM2-24B-A2B, agentic and tool-use benchmarks against Qwen3.6-35B-A3B, inference speed on Apple and NVIDIA hardware, and how to get started.
6 min · 1,330 words
Frontier Models for Every Robot: Announcing our API beta
Physical Intelligence opens a beta of the Pi API, letting robotics developers upload data, fine-tune its latest robot foundation models on managed compute and run cloud-hosted inference on their own hardware, and shares results from early partners deploying it across different tasks, robots and industries.
3 min · 707 words
Indra OCR: a 1M-parameter OCR model that reads 49 pages per second on CPU
Datalab releases Indra OCR, an open OCR package built around a 1M-parameter recognition model that reads 49 pages per second on 32 CPU cores, 11-17x faster than Tesseract and PP-OCR, while returning full layout with line and block boxes, labels, reading order and clean markdown or JSON output.
7 min · 1,500 words
Announcing pg_vectorsearch: Scalable Open-Source Vector Search in PostgreSQL
Tiger Data launches pg_vectorsearch, an open-source PostgreSQL extension whose PRISM index uses partitioning and compression instead of HNSW graphs to deliver faster vector queries with far less memory and quicker index builds.
11 min · 2,613 words
Character.ai announces it has initiated a voluntary Chapter 11 bankruptcy process to facilitate a court-supervised sale, explaining the board’s reasoning, that the platform stays up during the process but will be wound down if no actionable bid emerges, and how users can manage or export their data.
2 min · 454 words
10 Years of the SRE book: Announcing the second edition of "Site Reliability Engineering"
Google announces a completely rewritten second edition of its Site Reliability Engineering book, ten years after the first: how SLOs, observability, automation, reliability testing, capacity management and SRE org models have matured, how SRE supports and uses AI, and why culture still matters.
3 min · 718 words
Announcing AWS Lambda Web Functions for fast, consistent web app performance anywhere
AWS introduces Lambda Web Functions: run standard HTTP servers (Express, Hono, Fastify, Next.js and other Node frameworks) on Lambda without event-handler adapters, with multi-concurrent requests, active CPU billing, one-command multi-region deploy and latency-based routing alongside Event Functions and MicroVMs.
5 min · 1,051 words
meshapplied lays out its view that time per task is the key AI metric and releases its first two quantised model-kernel combos: Goethe Qwen3.8-27B NVFP4 for consumer 32GB GPUs at about 255 tok/s, and Goethe Qwen3.8-Flash-Next NVFP4 for agentic coding at about 184 tok/s on a single 96GB card.
3 min · 718 words
LittleJS, the small dependency-free HTML5 game engine, gets its biggest update yet: a full 3D renderer that shares the canvas with 2D games, built-in 2D and 3D level editors, ready-made particle effects, a faster core, new tools for making games with AI, and a new demo site, alongside the third LittleJS Game Jam.
3 min · 765 words
SlashDocs: the platform behind 28k+ documents at Razorpay, now open source
Razorpay open-sources SlashDocs, the self-hosted MIT-licensed hub where AI agents publish interactive HTML docs and humans review them with anchored comments, versions, CLI, and MCP—powering 28k+ internal documents.
9 min · 2,045 words
Meta AI Research introduces the Muse Wonder series (Muse Wonder and Muse Wonder Full), frontier models aimed at personal and professional agentic tasks that combine Active Memory and Active Focus, available through the Meta Model API and Muse Code.
4 min · 1,030 words
PopVax founder Soham Sankaran announces that its broadly protective COVID-19 vaccine candidate PVX-001 entered a first-in-human Phase I trial in Melbourne, with 14 participants dosed and no serious adverse events as of 6 October 2026, and describes the company's full-stack approach from computational protein design to in-house mRNA manufacturing.
16 min · 3,593 words
Ending the Casuarina Linux Experiment
Wesley Moore announces he is winding down Casuarina Linux, the distro he started: things continue as-is until the end of October 2026, after which package updates stop, with infrastructure staying up until the end of 2026. He explains why he is ending the project and what happens to the package repo and website.
4 min · 831 words
Zerobrew: A faster alternative to Homebrew
Zerobrew is an open-source package manager positioned as a faster Homebrew alternative, documenting install flow, architecture, and claimed speedups up to about 6.6×.
8 min · 1,811 words
Microsoft-Decision-1: Our model for fast decision-making
Microsoft introduces Microsoft-Decision-1, a decision-scoring model in Foundry built to return structured outputs software can act on immediately, and reports that it leads on latency and quality across structured decision tasks against LLMs and other decision models.
6 min · 1,358 words
TypeSafe AI, the lab behind the Jev System One model, announces an $870 million Series A at a $7.5B valuation led by Andreessen Horowitz with Sequoia Capital and DCVC, with Martin Casado joining the board, and explains what the money means for developers and customers.
1 min · 213 words
A statement on the Tor Project's relationship with Mullvad
The Tor Project explains why it is continuing its technical collaboration with Mullvad after concerns over a political donation by a Mullvad co-founder, while pausing proactive co-branding and revising how Mullvad Browser is described on Tor's channels to avoid implied endorsement.
2 min · 538 words
Oxide Computer Company announces a $445M Series D, opening with the milestone that ordinary operations made it pay income tax this spring, and explains why a profitable hardware company still raises: to keep accepting new demand, expand manufacturing capacity and build an enduring company.
3 min · 694 words
Ryan Dahl announces that the entire Deno team is joining Cloudflare, explaining how the move continues Deno's goal of simpler server software beyond the runtime, what it means for Deno Deploy and the open-source runtime, and why the new celld work focuses on Durable Objects.
2 min · 512 words
$50K First Letters Prize awarded for PHerc 343
The Vesuvius Challenge awards a $50K First Letters Prize to Erwin Nieuwlaar for virtually unwrapping and detecting ink in PHerc. 343, revealing the first text read from this still-rolled Herculaneum scroll since AD 79. Papyrologists read a Greek sequence likely meaning 'everything that happens' and discuss possible references to music.
4 min · 975 words