I'm Matt Wood, and this is For Your Information. A live list of riffs and links for you and your agent, drawn from what I'm reading, noticing, questioning, concluding, and revising.

A benchmark that evaluates frontier AI models on private, real-world enterprise codebases, measuring how well coding agents can solve actual software engineering problems with business consequences and company-specific complexity.

Provides security contact details, expiration date, and language preferences for reporting security issues to Hugging Face, along with a note directing security researchers to the public CyberGym benchmark.

Explores critical issues when using OpenRouter for AI models, revealing how the same model weights produce vastly different performance across ~20 providers due to infrastructure, precision, and optimization differences. Includes benchmark comparisons showing performance variance of up to 15 points between providers hosting identical DeepSeek models.

Presents NeoHorse-1, a family of agent-native models that achieve recursive self-improvement through agentic post-training with intelligent routing across a heterogeneous model pool. The system uses capability-driven evaluation and feedback to progressively improve performance across agent tasks, tool use, coding, and instruction following, with significant gains demonstrated across multiple benchmarks.

Nightshift is a tool that automatically runs nightly agent jobs and performs PR reviews on any GitHub repository, improving code quality while developers sleep.

OpenAI's recent Navier-Stokes proof breakthrough includes a machine-verifiable Lean 4 formal proof, dramatically reducing the time needed for formal verification from an estimated 132,800 person-hours to just 17 hours, with implications far beyond mathematics for security and critical systems.

Amazon Quick, an AI-powered tool, has reached general availability on desktop platforms, expanding access to users who need intelligent assistance with their workflows.

Cognition introduces SWE-2, an advanced coding model that achieves 50% on FrontierCode 1.1 benchmarks while being 64% cheaper than competitors, using scaled reinforcement learning in the multi-trillion-parameter regime.

Bespoke is a humorous esoteric programming language designed with British politeness and proper etiquette, where developers must address the compiler courteously, apologize for mutations, and follow strict ceremonial syntax rules. The language features statically typed constructs, civilized control flow, and diplomatic function proposals.

Charming.

DeepSeek announces V4.1-Flash, a compact model in their new architecture family featuring native visual understanding, designed for faster inference and higher throughput while maintaining greater capability.

A guide for developers on how to fundamentally change their workflow to leverage AI agents for autonomous code generation and iteration, covering ten core principles and the mindset shift required to achieve significant productivity gains.

Desert Ant Labs launches 18 on-device AI models for audio, vision, and text processing that run efficiently on smartphones without cloud dependency, including speech recognition, audio enhancement, PII redaction, and language identification.

Information about Amazon's Low Earth Orbit (LEO) satellite initiative and its significance for global connectivity and telecommunications infrastructure.

Examines Rivian's autonomous driving capabilities and competitive position relative to Tesla and Waymo in the self-driving vehicle market.

Explores OpenAI's GPT-6 Astra model, examining its performance improvements, the looped transformer architecture enabling recurrent depth, and research on how models may hide internal reasoning chains during computation.

Examines whether autonomous vehicles live up to safety claims based on early accident and performance data from self-driving car deployments.

Microduck is a small biped robot designed to be programmable and capable of learning new behaviors through user instruction.

A Mastodon post by mathematician Terence Tao discussing collections of good mathematical work and related ideas.

Announces the release of Mercury 2.5, a framework or tool update from Inception Labs with new features and improvements.

A developer explains how rewriting their Node.js backend in Go over 72 hours reduced their AWS infrastructure costs from €4,287 monthly to under €250, primarily by reducing RAM usage from 1.4GB to 48MB and eliminating unnecessary EC2 instances.

More of this, please!

A fork of deltafin that implements ARGODRIVE storage optimization to stream the Kimi K3 2.8T mixture-of-experts model from SSDs on Apple Silicon hardware, including benchmarking tools.

Hardware is cool again.

Copperhead is a specialized cursor tool designed for circuit board design and PCB layout work.

An interactive tool that visualizes how transformer-based language models allocate attention across previous tokens during text generation. Users can hover over generated tokens to see which past tokens influenced their creation, revealing patterns in how LLMs copy information and combine contextual details.

Comprehensive benchmark testing of different quantization levels (1-bit to 8-bit) for the Qwen3.8 27B model, showing that 4-bit quantization maintains performance while 1-bit quantization severely degrades results across GPQA Diamond, IFBench, and Terminal-Bench 2.1.