Anthropic launches the Model Hardware Standard (MHS), a universal USB-C-like interface enabling AI agents to seamlessly operate physical machinery across industries like robotics and manufacturing, with plans to open-source the technology as competition with OpenAI and Amazon intensifies.
Nvidia is reportedly acquiring AI model hub Hugging Face for $12.9 billion, a move that could give the chipmaker dominant control over the open-weight AI ecosystem while countering rivals OpenAI and Anthropic's push into custom hardware.
Alibaba unveils Qwen3.8-Flash-Next, a preview of its next-gen Qwen4 architecture boasting 125B parameters but activating only 6B per token, slashing compute costs by roughly a third while cleverly sidestepping US chip export restrictions — though its commercial licensing terms may put European users at legal risk under the EU AI Act.
Tencent's WeChat Vision Team releases WeMM-Embedding, a family of multimodal AI models in 2B, 4B, and 9B sizes that claim top scores on MMEB-v2 and MMEB-v3 benchmarks, supporting text, images, videos, and documents, now publicly available on Hugging Face under Apache License 2.0.
A viral open-source tool called claude-obsidian is taking GitHub by storm with 13.5K stars, transforming Claude Code into a self-organizing AI knowledge base that automatically links and files any source into a fully user-owned, local-first Markdown graph with zero silent cloud uploads.
IBM launches Granite 4.2, a family of open-weight AI models in 3B, 8B, and 30B variants featuring agentic capabilities like web search and tool integration, directly challenging OpenAI and Anthropic as enterprises increasingly seek cost-effective, self-hosted alternatives to expensive cloud-based frontier models.
FreeToken is revolutionizing local AI by enabling anyone with a gaming PC to run massive 290B+ parameter AI models at interactive speeds, using an open-source inference engine with smart CPU-GPU co-execution and semantic caching — no expensive cloud subscription required.
JD.com launches JoyAI-Echo, an open-source AI platform featuring two groundbreaking tools: Echo-LongVideo, capable of generating up to 5-minute multi-shot audio-visual content, and Echo-WM, an omnimodal world model that simultaneously evolves video, sound, music, and speech in response to real-time navigation inputs.
FastVideo, a new open-source framework, delivers a groundbreaking 50x video generation speedup across NVIDIA GPUs, Apple Silicon, and cloud platforms, featuring sparse distillation, LoRA finetuning, and real-time inference capabilities, with recent updates adding synchronized video-audio generation and Apple Silicon-optimized models.
Marker, a powerful open-source conversion tool, transforms PDFs, DOCX, XLSX, and more into markdown or JSON at industry-leading speeds of up to 23.7 pages per second, outperforming rivals MinerU and docling with a 76% accuracy benchmark score while supporting optional AI integration via Gemini, Claude, and OpenAI.