AI Briefing
TLDR AI news
The latest AI and developer news from TLDR AI, with short summaries.
Visit website →How Microsoft Deploys Thousands of Production AI Agents
This covers the technical challenges and solutions Microsoft encountered while operating enterprise-scale AI agents through Microsoft Foundry.
Mantis Skills: A Portable Toolkit for Building Security Review Harnesses (GitHub Repo)
Mantis Skills, a flexible and extensible toolkit designed for security review of coding agents, has been released.
Agent Benchmark Testing Long-Horizon Terminal Task Performance (GitHub Repo)
The LHTB benchmark has been released to measure whether LLM agents can perform complex tasks spanning hundreds of steps in containerized terminal environments.
Video Generation Models as General-Purpose Vision Models
GenCeption is proposed as a general-purpose vision model that performs diverse visual cognition tasks by leveraging video generation models.
How to Build Fable, Cheaper Than Opus
Fable 5 has a higher per-token cost than Opus 4.8, but through efficient management it lowers overall agent operating costs while improving performance.
Smart Cellular Bricks: Collective Intelligence for the Physical World
Sakana AI has unveiled a distributed hardware system that recognizes its own shape and detects damage without any central control.
Today, we're launching verifiers v1 (3 min read)
Prime Intellect has launched verifiers v1, an environment stack for agentic RL and evals.
Better Call Sol: A Core Tool for Practical Work
GPT-5.6 Sol is praised as a default model for advanced knowledge work due to its excellent balance of reasoning quality, speed, and cost.
Open-Weight Models Account for 29% of AI Gateway Usage
According to Vercel data, Open-Weight models accounted for 29% of AI Gateway token usage while showing high efficiency.
OpenAI Temporarily Eases Usage Limits for GPT-5.6 Sol
In response to a surge in demand, OpenAI has temporarily lifted the 5-hour usage limit for GPT-5.6 Sol and reset usage counts.
Cursor Developing General-Purpose AI Agent to Rival Anthropic's Claude Computer Use
AI code editor Cursor is developing a general-purpose AI agent called 'Sand' that goes beyond coding to handle office tasks.
Power Consolidates Around Greg Brockman at OpenAI Ahead of IPO
OpenAI's Fidji Simo is stepping back due to health issues, leaving President Greg Brockman to take over product and business operations across the board.
OpenWiki Brains: Proactive Memory for AI Agents
OpenWiki Brains is a proactive memory framework that autonomously gathers information from various sources to provide AI agents with persistent context.
Claude Code for Desktop Adds In-App Browser Feature
Claude Code for Desktop now supports an in-app browser feature for checking documents and designs.
Anthropic Extends Claude Fable 5 Availability Period (2-Minute Read)
Due to computing resource constraints, Anthropic has extended the availability period of Claude Fable 5's paid plan to July 19.
Meta Withdraws New AI Image Feature Amid Fierce Backlash
Meta withdrew an AI image generation feature that used public account content due to privacy concerns.
Apple Sues OpenAI
Apple has filed a lawsuit against OpenAI and former executives, alleging leaks of hardware confidential information.
GPT-5.6: Frontier Intelligence That Scales to Your Ambition (19 min read)
OpenAI released GPT-5.6, including the Sol, Terra, and Luna models, showcasing multi-agent processing and enhanced design judgment capabilities.
Tencent in talks to acquire Manus stake after Meta's acquisition falls through
Tencent is negotiating to acquire a stake in Manus, valued at $2 billion, that opened up after Meta's acquisition fell through.
Meta to Begin Mass Production of New AI Chip Starting in September
Meta will begin mass-producing the next-generation version of its self-developed AI chip, MTIA, starting in September to reduce GPU costs.