NVIDIA has released a 2.6B open-source world model for 1-minute 720p video.
Open-weight LLMs including Gemma 4 and DeepSeek V4 are lowering long-context costs through KV sharing and compressed attention.
An MCP server that diagnoses citation potential in AI search engines and optimizes pages has been released.
Fixed-size online memory improves LLM long-term memory performance.
Codex turns a data science team's scattered inputs into a reviewable first draft of analysis.
Codex organizes scattered sales context into drafts ready for action.
The UK Parliament is reviewing an amendment that would allow data centres and AI systems to be shut down in an AI emergency.
This analyzes the components of the GGUF file format—chat templates, special tokens, sampler settings—and the areas that still need improvement.
Anthropic has released a bundle of Claude plugins and agents for legal practice.
Claude Code has organized its operating principles for large codebases into an official guide.
A compilation of 10 months of Anthropic Claude limit changes from official sources.
It's an MCP-based plugin that delegates heavy code generation tasks to a local LLM to reduce token consumption in Claude Code and Codex CLI.
Mandala, an open-source runtime that streams logistics data into OpenTelemetry spans so AI agents can reason over it, has been released.
OpenAI acquired Weights.gg and shut down its unauthorized celebrity voice catalog.
In CC 2.1.143, the impossibility-condition judgment and Verify skill verification rules have been strengthened.
This introduces Superpowers, a Claude Code plugin that enforces a planning phase during AI coding to prevent reckless, unplanned code generation.
Google Chrome is automatically installing the Gemini Nano model in the background without user consent.
The article argues that portability in AI kernels is a myth, and top performance comes from hardware-specific DSLs.
Malta has partnered with OpenAI to become the first country in the world to offer all citizens ChatGPT Plus and AI education free for one year.
Orthrus, based on Qwen3, achieved up to 7.8x speedup through lossless parallel generation.
Meta will use mouse and keystroke inputs and screen captures from U.S. employees' work PCs to train AI.
Stanford found that agentic AI increased productivity by 71% across 51 enterprise case studies.
Ontario's Auditor General audited 20 AI Scribe systems and found widespread errors.
OpenAI has launched a feature that integrates with Plaid to enable personal finance management within ChatGPT.