LG AI Research has unveiled Cluster-Aware Upcycling technology that reduces expert redundancy and maximizes specialization when converting a Dense model into an MoE.
A developer used a Claude-powered automated pipeline to process 27 issues over two weeks for $110 a month.
ReactBench v1 comprehensively evaluates coding agents' React-writing ability using the React Doctor validator, which has over 400 rules.
Goodfire unveiled the beta of its Silico platform, which automates AI research experiments.
OpenAI and Work Louder unveiled Codex Micro, a mechanical keyboard for AI agent work, priced at $230.
Vercel has released AI Gateway leaderboard data under a CC BY 4.0 license.
This analysis examines the design principles and efficiency of the Systolic Array, which accounts for over 95% of the performance of modern AI chips.
NVIDIA unveiled the Jetson T3000/T2000, compact high-performance modules for robots and edge AI.
Open Interpreter is a coding agent optimized for low-cost AI models, delivering top performance through a Kimi K3 Rust implementation.
Using GPT-Red, an automated security tester developed by OpenAI, vulnerabilities in AI models were found and the safety of GPT-5.6 was improved by 6x.
SpaceXAI has open-sourced Grok Build, a terminal-based AI coding agent.
OpenAI's GPT-5.6 Sol achieved first place on the Design Arena benchmark for the first time, showing a new approach that deliberately avoids AI design anti-patterns.
The AIDE² system surpassed a model hand-tuned over 2 years through 8 days of recursive self-improvement.
Model routing in AI agent systems isn't a model-selection problem but a system optimization problem involving caching, infrastructure, and regulations.
Anthropic is pursuing an IPO this October, having surpassed OpenAI with a $96.5 billion valuation.
Thinking Machines has unveiled Inkling, a 975B-parameter multimodal open-weight model.
LLMs can significantly improve their code generation ability using only their own outputs, without needing a verifier or separate training.
Apple TV combined Text and ID Embeddings for personalized search, improving accuracy by 8% on short queries.
Grok has launched an automation feature that runs tasks automatically on a schedule or via email triggers once a user describes the task a single time.
A write utilization metric has been added to Vercel ISR Observability, making it possible to identify inefficient revalidations.
Cars24 recovered 12% of lost leads while handling 1 million minutes of customer conversations per month with OpenAI agents.
Moonshot AI's Kimi K3 has become available on Vercel AI Gateway.
We built a doubly sub-linear interactive proof of proximity (dsIPPs) scheme that can verify properties of data by reading only a tiny fraction of the input data.
They built an interactive proof system that can efficiently verify complex properties of data distributions using limited resources.
AI Briefing