MoBA, a technology that applies the MoE principle to attention to maximize the efficiency of long-context processing, has been announced.
Cerebras unveiled the architecture of a high-performance RAG system built by integrating diverse data sources.
Stoke, an open-source tool that supports efficient routing and management between Ollama instances running on multiple machines, has been released.
Cursor revealed the results of an experiment showing that an Agent Swarm structure improves the efficiency and economics of complex coding tasks.
LangChain built IssueBench, a synthetic benchmark for evaluating LangSmith Engine's ability to identify and categorize issues.
Kimi Work, a desktop AI agent supporting local file integration and web browsing automation, has been launched.
AWS announced a range of updates including Lambda setup prompts for coding agents and support for OpenAI GPT-5.6 models on Amazon Bedrock.
About 32% of newly submitted arXiv papers were identified as containing AI-written text, with computer science showing the highest rate.
Starting September 1, 2026, Anthropic will uniformly raise the API pricing for Claude Sonnet 5 by 50%.
The base model of DWARF, a new Sparse Attention architecture aiming for O(1) complexity, has been released.
aiignore, an open spec that defines the scope of data an AI agent can access or modify, has been released.
NVIDIA has released Cosmos 3 Edge, a 4B-parameter world model capable of real-time inference on edge devices.
As the transparency provisions of the EU AI Act take effect, labeling obligations for AI-generated content are strengthened.
As enterprise AI agents proliferate, the importance of LLM gateways for managing cost, control, and compliance is being emphasized.
This describes how Dropbox optimized and evolved Riviera, its content processing platform developed over 10 years, for AI.
Unsloth now officially supports local LLM inference and fine-tuning using AMD GPUs.
This covers cases of discovering security vulnerabilities using AI, tackling mathematical challenges, and bypassing security constraints.
Trellis.cpp has launched a Studio feature that supports automatic model downloads and 3D preview.
OpenBMB has released MiniCPM5-2B, a model that recorded top-tier performance among 4B-class models.
Israel is conducting a large-scale AI influence operation that manipulates content on the web to steer LLM answers in a specific direction.
Inertia-1, a motion foundation model universally applicable across diverse sensors and body parts, has been released.
A problem has been reported where Subagent logs abnormally accumulate in resumed long-running Codex CLI sessions, exhausting disk space.
The US administration is considering a de facto ban on Chinese open-source AI models, tightening regulations.
This summarizes the concept and latest research on Diffusion Language Models (dLLMs), which aim to overcome the limitations of the autoregressive approach.
AI Briefing