Blog
The pieces that do not fit in a glossary entry: where a benchmark number came from, what a claim actually shows, and what something costs in production.
NVIDIA’s $12.9B Hugging Face Deal Is a Bet on Open Models
NVIDIA will acquire Hugging Face for $12.9 billion. Why the deal's incentives favor open models — and where the neutrality risk actually sits.
Meta Ships Muse Spark 1.3 With Big Long-Context Gains
Muse Spark 1.3 posts 98.5 on MRCR and cuts tokens 25% on coding — but cost per task rose 37%, and the headline scores use a mode that has not shipped.
First Complete Male Fruit Fly Connectome Mapped With AI
Google Research, HHMI Janelia, MRC LMB and Cambridge published the first full wiring diagram of a male fruit fly nervous system: 166,691 neurons, 11,691 types.
Inworld Ships Realtime TTS-2 Voice Model Family
Inworld completes its Realtime TTS-2 rollout with a Flash variant and per-character pricing. Where the model actually ranks on the Artificial Analysis arena.
Dyson CameraJet: A $499 Toothbrush That Runs Computer Vision
Dyson's CameraJet puts a 100,000-pixel camera in a toothbrush head and uses on-device machine vision to spot gaps between teeth and jet them clean.
DeepSeek Plans a 160,000-Chip Huawei Cluster in Inner Mongolia
Bloomberg reports DeepSeek will deploy at least 160,000 Huawei Ascend 950DT accelerators in Inner Mongolia — for inference. Training stays on Nvidia.
Anthropic Ships Open-Source Commerce Agent Blueprints
Anthropic released Apache-2.0 blueprints for a shopping agent and a merchant agent on Claude, with Visa, Mastercard and Accenture as distribution partners.
Alibaba’s Qoder Coding Agent Gets a Smart Glasses Client
Alibaba announced a glasses edition of its Qoder coding platform on September 3, 2026, running on Qwen and Rokid AI glasses with full-duplex voice.
World Labs Announces Atlas, an Omni World Model
World Labs unveiled Atlas on September 1, 2026: an omni world model over text, images, video and 3D that generates a minute of 1440p camera-controlled video.
Gemini Now Picks Which Parts of a Video to Watch
Google's agentic video understanding lets Gemini load video segments on demand. An hour of footage drops from ~475K tokens to a fraction of that.