Very proud to share what we've been working on this year. Self-improvement has been a long, ambitious dream. It's incredible to see early success in such complex, 3D, embodied environments. Let's go! 🚀
SIMA 2 is our most capable AI agent for virtual 3D worlds. 👾🌐
Powered by Gemini, it goes beyond following basic instructions to think, understand, and take actions in interactive environments – meaning you can talk to it through text, voice, or even images. Here’s how 🧵
@Steve_Yegge Maybe tell your buddy to do some actual work and to stop spreading absolute nonsense. This post is completely false and just pure clickbait.
2004 was a good year, but your Gmail address doesn't need to be stuck in it.
To say goodbye to [email protected] or [email protected] (or whatever you were into at the time), go to your Google Account settings and choose any name available. You'll keep your old username and you can sign in with both.
As timelines tighten, details might benefit attackers more than defenders. So we're trying something weird: proving a circuit exists without revealing it.
For example, here's a zero-knowledge proof that we found 10x smaller quantum circuits for ECDLP: https://t.co/ypwPEVurg5
wow
A recipe for endless self-improvement toward Embodied AGI is here
Genie 3 generates the world.
Gemini (Task Setter) makes a task.
SIMA 2 acts.
Gemini (Reward Model) scores the video.
"The agent continuously generates its own experience and learns from it" 👀
Practical humanoid robots are coming
This was the dream when we started the Genie project in 2022. At the time it seemed crazy to many, but to us it felt like the fastest path to building a truly general real world environment generator. Luckily some others at GDM believed in it and now we’re here!! 🧞🤯🚀🚀🚀
A recipe for endless self-improvement toward Embodied AGI: 🧬🤖
1️⃣ Genie 3 generates the world.
2️⃣ Gemini (Task Setter) makes a task.
3️⃣ SIMA 2 acts.
4️⃣ Gemini (Reward Model) scores the video.
The agent continuously generates its own experience and learns from it!
[1/N]
Can AI self-improve on its own and reach superhuman performance? 🧠
In our Sima 2 paper, we dropped a Gemini agent into an unseen 3D world. The model acted as the task proposer, the agent, and the reward model - autonomously learning from self-generated experience. It surpassed human performance through self-improvement iterations!
Actually if you want to know what the real ‘secret’ is 😀 it’s world-class research AND world-class engineering AND world-class infra all working closely together with relentless focus and intensity…
You went 🍌🍌 for Nano Banana. Now, meet Nano Banana Pro.
It’s SOTA for image generation + editing with more advanced world knowledge, text rendering, precision + controls. Built on Gemini 3, it’s really good at complex infographics - much like how engineers see the world:)
Gemini 3 is here. 🚀
It brings next-level strength in reasoning and coding, plus major upgrades in science and maths.
Extremely excited to see what you build with it.
A super model from our super team!
Its been an exciting 7 days of shipping
- New much improved Gemini Live on Android and iOS
- Gemini 3.0 Pro in Gemini App and AI Studio
- Search AI Mode with Gemini 3.0 Pro and much improved shopping experience
- Google Antigravity, our next generation agentic IDE
-- Nanobanana in Google Photos
- SIMA 2 research
- Waymo now across SF Bay Area and in Miami today
More to come!
This is Gemini 3: our most intelligent model that helps you learn, build and plan anything.
It comes with state-of-the-art reasoning capabilities, world-leading multimodal understanding, and enables new agentic coding experiences. 🧵