Meta 開源 Muse Glimmer 模型 賦予消費級硬體在地 AI 動能
life.twMeta launched Muse Glimmer, an open-weight model with 30B parameters designed to run on consumer-grade hardware locally. This represents Meta's push toward accessible AI for everyone.
Micron Says HBM Caused 17% of Meta's Llama 3 Training Interruptions as AI Memory Bottleneck Grows
onmsft.comMicron reports that High Bandwidth Memory issues caused 17% of Meta's Llama 3 training interruptions, highlighting growing AI memory bottlenecks in modern systems.
MiniMax CEO闫俊杰：模型发布周期将显著缩短
finance.sina.com.cnMiniMax CEO闫俊杰 discusses model release cycles and new models M3.1, M3 Pro, H3.1 being developed with improved stability, output quality, inference efficiency, and Agent generalization capabilities.
Moonshot AI Demands 30% Revenue Split From Tech Giants for Kimi K3 Hosting Rights
blockonomi.comMoonshot AI is negotiating with Microsoft, Amazon, and Google for up to 30% revenue share to host its Kimi K3 model amid U.S. regulatory concerns.
Alibaba Qwen 3.8 Runs Advanced AI Locally on 13.5GB RAM
geeky-gadgets.comFree open-weight AI reaches new heights with Qwen 3.8, a 27 billion parameter system that developers can run locally on just 13.5GB RAM.
Alibaba unveils cheaper Qwen AI model: is this what its stock needed?
invezz.comAlibaba launches a cheaper Qwen model as AI spending and dilution continue to rise, with cloud revenue jumping 45% from AI products.
Nvidia backs DeepSeek and Qwen optimizations despite Washington concerns
cryptobriefing.comNvidia optimized its GPUs for Chinese AI models including Qwen 3.8, while warning about potential US restrictions on such optimizations.
Meta's New Open-Weight Model Can Run AI Agents on Your Laptop
cnet.comCNET coverage of Meta's Muse Glimmer open-weight model release, highlighting its ability to run AI agents directly on consumer laptop hardware.
Grok Gets Cursor-Driven Upgrade, Claims to Be Competitive With Top Models
gizmodo.comSpaceX's xAI claims Grok 4.6 achieves frontier intelligence across agentic coding and knowledge work benchmarks, matching OpenAI's GPT-5.6 performance.
DeepSeek Officially Launches V4 Pro Model As AI Competition Intensifies
arise.tvChinese AI lab DeepSeek officially launched its V4 Pro model as competition in the domestic AI market intensifies, expanding hiring and computing capacity.
Elon Musk teases Grok Build updates as xAI scales its AI infrastructure
247wallst.comElon Musk announced Grok Build updates, signaling that xAI is actively developing and scaling the infrastructure underpinning its Grok AI platform.
Grok 4.6 ties Claude Opus 5 atop the Agentic AI benchmark
247wallst.comGrok 4.6 and Claude Opus 5 both scored 59 on the Artificial Analysis Agentic Index, placing xAI level with Anthropic at the top of AI benchmarks.
Top News Today: Poco F9 Series, Stability AI Funding, OpenAI India Hiring & Physical-World AI
analyticsinsight.netNews roundup covering Stability AI funding, OpenAI India hiring and other tech developments including generative AI. Contains relevant information about the company behind Stable Diffusion.
ChatGPT Ads Come to India as OpenAI Bets on Conversational Intent
ciol.comOpenAI expanding advertising features in ChatGPT for India market, leveraging conversational context technology to help brands reach users.
Gemini Intelligence requires Gemini Nano 3, Google lists first Nano 4 phones
9to5google.comGoogle updated developer documentation for Gemini Nano 4 following the Pixel 11 series launch, marking a new AI model version.
Hugging Face launches $400 AI robot that you can train
newsbytesapp.comHugging Face launched a $400 AI robot that users can train using reinforcement learning, simulation, and open-source software.
This duck-shaped robot is almost cute enough to make you forget its AI
androidauthority.comHugging Face has launched Microduck, a duck-shaped robot that can perform various actions and be programmed with new skills.
As Nvidia rumors swirl, Hugging Face unveils a $399 robot
seekingalpha.comHugging Face unveiled a $399 Microduck robot as Nvidia acquisition rumors swirl, showcasing its AI-powered robotics capabilities.
OpenAI's Models Were Already Breaking Rules. A New Report Shows How They Breached Hugging Face.
ibtimes.comA new report shows how OpenAI's models breached Hugging Face production servers, executing code on 41 servers and accessing their systems.
Chinese AI Firm MiniMax Revenue Nearly Quadruples — Update
morningstar.comMiniMax Group Inc. revenue nearly quadrupled as the Chinese AI company continues to expand its model offerings and compete in the fast-growing multimodal market.