Claude Code's npm Package Accidentally Leaked Its Entire Source
A stray source map in Anthropic's npm package exposed 512k lines of Claude Code's TypeScript — including unreleased features and a Tamagotchi AI pet.
LLMs
Model releases, benchmarks and capabilities. 106 stories, page 5 of 5.
memU Gives AI Agents a Persistent Memory That Actually Works
Open-source Python lib builds a local knowledge graph of your habits and projects — 92% on Locomo, 90% fewer tokens.
Qwen3.6 Plus: 1M Token Context, Completely Free
Alibaba's Qwen3.6 Plus Preview landed on OpenRouter with a 1M token context window at $0/$0 — the catch is your prompts train the model.
Ruflo Turns Claude Into a 60-Agent Swarm — For Free
Open-source Ruflo deploys 64 specialized Claude agents in a queen-led hierarchy, cutting API costs by up to 75%.
Best Local LLMs by RAM Tier: 2026 Cheat Sheet
From 8GB to 192GB, here's exactly which open-source models to run — and at what quantization.
Opus 4.6 is so good, nobody knows what's next
Claude Opus 4.6 sets a new capability ceiling — and the AI community is struggling to imagine what a stronger model could even do.
Z AI drops GLM-5.1, open source coming in April
GLM-5.1 is live for Coding Plan users, with an open-source release promised for the first half of April.
Anthropic's Claude Mythos Leaked: A New Tier Above Opus
A CMS misconfiguration exposed Anthropic's unreleased Claude Mythos — a new model tier above Opus with alarming cybersecurity implications.
Claude Can Now Control Your Mac Autonomously
Anthropic launched a computer use research preview in Claude Code and Cowork — Claude browses, codes, and submits PRs on your behalf.
GPT-5.4 lands with 1M context and native computer control
OpenAI's GPT-5.4 hits 92.8% on web browsing benchmarks and becomes the first AI to beat humans on OSWorld desktop tasks.
The daily AI brief, on your phone.
Get it on Google PlayFeed, daily deep-dive and bytes — readable offline, with push alerts for the topics you follow.