Google dropped MTP versions of Gemma4. Ran them on my DGX Spark. The 31B dense model went from 3.94 → 8.91 tok/s. That's +126%. Full results: [26B A4B] > 25.24 → 31.69 tok/s (+25.6%) > TTFT 755 → 332ms (-56%) [31B] > 3.94 → 8.91 tok/s (+126%) > TTFT 599 → 378ms (-37%) If you're not running MTP, you're leaving free perf on the table.

Google, Microsoft and xAI will share unreleased versions of their AI models with the government to curb cybersecurity threats, the National Institute of Standards and Technology announced. https://t.co/vKToPCOh6m https://t.co/jX6X9uMwCu

Google cofounder Sergey Brin’s $42 million purchase, along with a record $125 million sale in Incline Village, marks a new era for Tahoe real estate as billionaire buyers flock to the Nevada side. Read more: https://t.co/p6MvA5XR2U Photo: Realtor .com https://t.co/mUVLPCnnuS

Exclusive: Google DeepMind will train its AI technology on EVE Online after Google took a multi-million-dollar stake in the sci-fi MMORPG's developer. EVE Online is famous for players' corporate espionage, economic maneuvering and politicking. https://t.co/5P6nWZqIjL

NEWS: Google has signed a classified deal with the US Department of Defense to provide AI models for lawful government purposes, joining OpenAI and xAI in a move that follows the blacklisting of Anthropic from federal use. The agreement, according to The Information, allows the Pentagon to use Google's models for various government purposes but does not grant the company veto power over operational decisions. https://t.co/EdtvuFXQLE

NEWS: Microsoft, Google, and Elon Musk's xAI have agreed to provide the U.S. government with early access to new AI models for national security testing, following growing concerns over the hacking capabilities of advanced systems like Anthropic's Mythos. The agreement allows the Department of Commerce's Center for AI Standards and Innovation (CAISI) to evaluate models for security risks and unexpected behaviors before they are widely deployed to the public. This move expands on previous agreements with OpenAI and Anthropic to establish a framework for vetting frontier AI models for national security implications. CAISI Director Chris Fall stated that independent, rigorous measurement science is essential to understanding the implications of frontier AI. The center has already completed over 40 evaluations, including testing on state-of-the-art models that have not yet been released to the public. https://t.co/CFbJXM5BU5