saw a lot of posts about the Dream-RSI paper published by Google DeepMind & collaborators. From my (basic) understanding, they've developed a way to analyze past runs (like a video game) to discover the optimal speedrun route.
tl;dr - you finish Level 1-1, then review your run
[first date]
him: what do u do
me: [remembering men like ambitious women] i'm building AGI
him: are u serious
me: [remembering men like humility] it's a wrapper
Try the new Gemini 3.8 Live models when making any agents or bots that talk! If you want to create an agent that you can practice speaking a new language in, practice your speech, or any speaking feedback here is a brand new model for you :)
at this point I just assume all my data has been leaked onto the darkweb for $2 or less... I have 2fa, password manager, passkey, and my credit frozen, and hope for the best 🥲
I really recommending reading this.
In summary, a company which does ID verification for in-person interactions (hotels, car rentals, ID verification for alcohol or marijuana, etc) has some how exposed over 153,000,000 drivers licenses for people in the United States and Canada.
the benchmarks for 3.8 flash are pretty crazy...!!
🗣️🗞️ come get your state of the art frontier model at FLASH tier pricing! rolling out now on antigravity and Gemini users :-)
Introducing Gemini 3.8, our best reasoning & coding model yet.
By leveraging long-running agentic loops, we’re building on the momentum of 3.7 Flash from just three weeks ago to release two new 3.8 variants:
Meet Gemini 3.8 Flash and Gemini 3.8 Flash Cyber 🧵
