The new model is up to twice as successful at some tasks.
DeepMind reveals Gemini Robotics 2.5 and open safety benchmark "ASIMOV-jailbreak"
Source: Ars Technica
Original reporting
Ars Technica: read the full reportRelated stories
TechCrunch
Apple TV makes sign language interpretation a built-in accessibility feature
Apple TV's new accessibility feature lets viewers turn on sign language interpretation within supported titles.
The Verge
The Verge’s 2026 holiday gift guide for moms
Given everything moms do for us, it feels good to give back during the holidays. But where to start? How do you come up with a physical representation of how you feel about such a key and influential figure in your life? The possibilities seem as endless as they are daunting, but fear not, we've […]
The Verge
Google releases Gemini 4 Carbon with agent ‘swarm’ feature
Google is rolling out its latest AI model, Gemini 4 Carbon, to the general public after releasing it to a select group of "cybersecurity defenders" two weeks ago. Google claims Carbon is its "strongest model" yet, and capable of competing with the very best systems from Anthropic and OpenAI. Google says Gemini 4 Carbon "pushes […]
TechCrunch
Google broadens its AI hardware portfolio with two new specialized chips
Google's new tensor processing units (TPUs) broaden its chip portfolio to more use cases: confidential model training and more memory capacity.
The Verge
Ember’s latest self-heating 14-ounce mug is $50 off
If you were cool on the $149.95 original price of the 14-ounce Ember Mug 3, you might warm up to the current discount happening at Amazon, Ember, and Best Buy for several colors of the (top rack) dishwasher-safe mug. It’s 33 percent off, dropping to $99.99 — still not cheap, but cheaper, and notably just […]
TechCrunch
N-able takes its remote access service offline after researcher publishes dozens of zero-day bugs
The remote access giant also urged its self-hosted customers to take down their own servers amid a "critical security incident," fearing the possibility of cyberattacks.