The daily ledger

🦞 LIVE NOW: Build a personal AI agent you own

How to Use OpenCode for Free: Beginner's Desktop Setup Guide

OpenCode gives you a full AI coding agent on the desktop, plus rotating free models, a paid subscription, pay-as-you-go access, and support for your own AI providers. Here's how to get started — and what you're actually…

xAI’s Grok 4.7 Makes Frontier AI a Price War

xAI’s Grok 4.7 isn’t clearly the smartest AI model available. That may not matter. By getting closer to GPT-6 Astra and Claude Fable 5.1 while charging dramatically less, xAI is turning the frontier model race into a…

A New Kind of Model for AI Decision-Making?

Exploring TypeSafe's Jev and comparing it with OpenAI on intent classification The post A New Kind of Model for AI Decision-Making? appeared first on Towards Data Science.

GPT-6 Astra Just Hit OpenAI's Highest Cybersecurity Risk Level

Why this matters less for what Astra can do and more for what every other model hasn't been tested for The post GPT-6 Astra Just Hit OpenAI's Highest Cybersecurity Risk Level appeared first on Towards Data Science.

Architect Labs Wants AI Chips to Evolve as Fast as AI Models

Architect Labs says its AI system turned a high-level specification into a working FPGA accelerator in less than two weeks. Redwood tests whether AI can shorten chip design cycles enough for specialized hardware to…

Google Offered $10M for a Dying Airline's Data. How Can You Value Yours?

How to put a dollar value on your organization's operational data The post Google Offered $10M for a Dying Airline's Data. How Can You Value Yours? appeared first on Towards Data Science.

Your AI Assistant Wrote the Code. Who Checked the Defaults?

Five scikit-learn defaults that deserve a closer look before your next model reaches production The post Your AI Assistant Wrote the Code. Who Checked the Defaults? appeared first on Towards Data Science.

Jensen Huang Says AI Won’t End the World. His Bigger Bet Is on Existing Law.

Jensen Huang says AI extinction by 2030 has a 0% chance of happening — and argues existing laws already give AI companies reason to build safely. Anthropic’s latest misuse cases turn that claim into a more immediate…

😸 Would GPT-6 stab a doll?

When AI Gets Arms, “Just Say No” Stops Being a Safety System

RoboHarm connected frontier AI models to real robot arms and asked them to perform deliberately hazardous tasks. The results reveal a bigger problem than unreliable refusals: physical AI needs independent controls that…

GUARD: Natural Forgetting in Large Reasoning Models via Guided Answer-Reasoning Distillation

arXiv:2609.21677v1 Announce Type: new Abstract: Recent advances in large reasoning models (LRMs) have made machine unlearning more challenging, as protected facts or unsafe rationales may surface in intermediate…

The Communication Bottleneck: A Round-Trip Study of Tree-Structured Expression Serialization in Language Models

arXiv:2609.21509v1 Announce Type: new Abstract: When language models reason in chain-of-thought or exchange free-text intermediates, they serialize structured information into natural language. How much tree-structured…

QuranicMMLU: A Cognitively-Aware Benchmark for Evaluating Generative AI Solutions on Quranic Linguistic Knowledge

arXiv:2609.22038v1 Announce Type: new Abstract: We introduce QuranicMMLU, a benchmark for evaluating generative AI on Quranic Arabic across multiple dimensions of linguistic complexity. Existing Quranic benchmarks…

From Task Success to Productive Success: Evaluating Human-AI Collaboration by Quality and Cost

arXiv:2609.21117v1 Announce Type: new Abstract: AI productivity is often measured by task completion time, economic value, or improvements in outcome quality. However, these measures usually treat collaboration as a…

Accelerating Dense LLMs via L0-regularized Mixture-of-Experts

arXiv:2609.21672v1 Announce Type: cross Abstract: Large language models (LLMs) achieve strong performance but suffer from slow and costly inference. Existing acceleration methods often lead to noticeable performance…

What Does Privileged Information Add to On-Policy Self-Distillation?

arXiv:2609.20612v2 Announce Type: replace Abstract: On-policy self-distillation (OPSD) lets a language model learn from a frozen copy of itself that sees an answer or a worked solution. Giving the teacher this extra…

Multi-turn Conversational AI from Text to Multimodal Interaction: Data, Models, Evaluation, and Open Challenges

arXiv:2608.17605v2 Announce Type: replace Abstract: Conversational AI is moving beyond isolated text prompts toward sustained, multimodal interaction. In real conversations, users clarify goals, revise requests,…

Do New Attention Mechanisms Actually Fix Attention Sinks at Million-Token Context?

arXiv:2609.08574v2 Announce Type: replace Abstract: Long context language models now advertise windows of one million tokens, but two habits limit how much of that window is used. Attention heads with nothing useful to…

Are You Sure You're Sure? On the Impact of Instruction Tuning on Confidence and Lexical Diversity

arXiv:2608.13430v2 Announce Type: replace Abstract: Instruction-tuned language models achieve strong performance across a range of generation tasks but have recently been shown to exhibit verbalized overconfidence,…