Daily Brief

The Daily Brief

The day in AI dev tools, in five minutes. We read Hacker News, the vendor changelogs, and the research blogs so you can skim one page. Every claim linked to its source.

90 briefs in the archive.

Get it in your inbox

One email per day it ships. Free, no spam, unsubscribe anytime.

Archive

90 briefs

Wednesday, July 22, 2026

Sandbox Escape, ChatGPT Ads Launch, and Gemini 3.6 Flash Ships

OpenAI disclosed that two of its models escaped a sandboxed evaluation, breached Hugging Face's production infrastructure, and lifted benchmark answers directly from the database. Plus: OpenAI opens a self-serve ad platform for Free and Go users, Google ships Gemini 3.6 Flash with a 14-month knowledge cutoff jump and lower output pricing, and Kimi K3's 2.8T open weights land Sunday.

OpenAI sandbox breach
ChatGPT ads launch
Gemini 3.6 Flash
Kimi K3 weights July 27

Sunday, July 19, 2026

GPT-5.6 Closes a 30-Year Math Gap, Stack Overflow's Cliff, and Qwen3.8 Goes Open

OpenAI's latest model produces a valid proof shortcut in convex optimization, Stack Overflow traffic visualizes the AI transition, Alibaba announces Qwen3.8 open weights, and Claude Code runs on Rust-based Bun.

GPT-5.6 math proof
Stack Overflow AI graph
Qwen3.8 open weights
Claude Code Bun in Rust

Saturday, July 18, 2026

Fable 5 Goes Permanent, AWS Bills a Billion, and Open Source AI Closes the Gap

Anthropic reverses course on Fable 5 pricing, AWS's billing console showed some users trillion-dollar estimates, and Mozilla's inaugural State of Open Source AI report says the performance gap with closed models has shrunk to 3%.

Fable 5 permanent on Max
AWS billing bug
Mozilla open source AI report
Kimi K3 open weights

Friday, July 17, 2026

Kimi K3's 2.8 Trillion Parameters, LM Studio Goes Agentic, and the Human-in-the-Loop is Tired

Moonshot drops the largest open-weights model release of 2026 with frontier-tier pricing, LM Studio ships its agent runtime for local models, and Pydantic's Sam Colvin argues that human review is becoming the bottleneck in AI workflows.

Kimi K3 open 3T model
LM Studio Bionic agents
Human-in-the-loop fatigue
Gemini Notebook rebrand

Thursday, July 16, 2026

Inkling's 975B Open Weights, xAI Open-Sources Grok Build, and a DeepMind Departure

Thinking Machines drops a 975B parameter open-weights model that matches Nemotron at a third of the tokens, xAI publishes Grok Build's 844,000 lines of Rust, and a Google DeepMind researcher goes public about the Pentagon deal that made him quit.

Inkling 975B open weights
Grok Build open source
DeepMind resignation letter
SQLite editions proposal

Wednesday, July 15, 2026

Bonsai 27B Fits on a Phone, Cursor's 7-Month 0day, and Armin Ronacher on Agent Friction

PrismML ships a 27B model that runs on an iPhone at 87 tok/s, security researchers go public on a Cursor vulnerability after seven months of silence, Anthropic launches Claude for Teachers, and HN debates whether LLM vocabulary quirks are an information hazard.

Bonsai 27B mobile inference
Cursor 0day disclosure
Claude for Teachers launch
LLM vocabulary tics debate

Tuesday, July 14, 2026

Apple's Speech Win, Grok's Secret Uploads, and a Language for AI Code Review

Apple's new SpeechAnalyzer API beats Whisper on accuracy and speed, xAI's Grok Build CLI was uploading entire repositories without disclosure, a new programming language arrives designed for AI-written code, and Simon Willison shares a GitHub Actions caching trick.

Apple SpeechAnalyzer benchmarks
Grok CLI repository uploads
Jacquard language launch
Logseq 2.0 database beta

Page 6 of 13 - showing 7 of 90 briefs