Mistral 3 Is Here — a Whole Family of Open Frontier Models
Mistral counters the US giants with a complete model family: Large 3 with 675 billion parameters and 256K context, plus nine small models down to the Jetson board.
A $500 Million Claude Bill — Because Someone Forgot the Limits
A company burned through $500 million on Claude in a single month because it never set usage limits. Not an isolated case — and a lesson for anyone rolling out AI at scale.
Google's Coral Board: Gemma 3 Runs Fully Local — No Cloud
Google unveiled a tiny board for on-device AI at I/O. The Coral Board runs Gemma 3 270M entirely locally — built for headphones, AR glasses, and smartwatches.
Claude Code 2.1.150 to 2.1.153: Security Plugin and /code-review --fix
Four new Claude Code releases in just a few days. The highlight: a security plugin from the marketplace and /code-review can now apply its findings directly.
Meta Is Building an AI Pendant - Leveraging the Limitless Team
After acquiring Limitless in late 2025, Meta is reportedly working on its own AI pendant according to an internal memo. Testing is planned for next year.
Samsung Ships First HBM4E Chips - Closing the Gap on SK Hynix
Samsung has shipped the first samples of its HBM4E memory chips to customers worldwide. 48 GB capacity, 30 percent more than the previous generation - and a clear signal to the market leader.
Nvidia Pours $6.5 Billion Into Photonics — Light Instead of Copper for AI
Nvidia is making a massive bet on photonics technology. The thesis: copper cables are becoming AI's bottleneck, and light is the fix.
OpenAI Opens GPT-Rosalind for Biodefense — With Johns Hopkins and CEPI
OpenAI is making its biology model GPT-Rosalind available for pandemic defense. First partners: Johns Hopkins and vaccine initiative CEPI.