Everything is video. We have said that on our last ICLR work openreview.net/pdf?id=w6YS9A7… 😁

I will be at ExHall D, Poster #198, with Mojtaba tomorrow morning (June 15), presenting Google Research 's newest multimodal generation model -- mmsr.kfmei.com. Come say hi! #CVPR2025

Context engineering is the future of prompt engineering -- presented in our recent CVPR25 work -- The Power of Context: How Multimodality Improves Image Super-Resolution mmsr.kfmei.com.

🔥 New work alert! We introduce KDS — an inference-time scaling method for diffusion models that reduces artifacts in image-to-image generation, inspired by the classic mean-shift algorithm. 🚗 Simple and training-free ⚡️ Compatible with most modern ODE-based solvers 🎓 Led

Mode seeking is all you need.

I joined the Veo team at DeepMind and am excited to be working on improving next-generation video models.

An advanced version of Gemini with Deep Think has officially achieved gold medal-level performance at the International Mathematical Olympiad. 🥇 It solved 5️⃣ out of 6️⃣ exceptionally difficult problems, involving algebra, combinatorics, geometry and number theory. Here’s how 🧵

To everyone in the NeurIPS rebuttal trenches: may your coffee be strong and your Reviewer 2 be kind. You got this!

Congrats to the launch!

I was so inspired by this exact series of early GAN works that I pivoted to video generation! This path led to our work on an early video diffusion model (VIDM) in 2022. arxiv.org/abs/2212.00235

magic banana

Big congratulation to the launch!

Finally nano banana is out!

Sherma's Song, Visualised 🎵 #silksong #hollowknight

Anyone who's done a PhD knows the feeling

Accepted by NeurIPS 2025! Congratulations to all colleagues and especially to the intern Yuyang Hu

Introducing Veo 3.1 📺! Proud to have been a core contributor to its diffusion sampling and post-training. After a period of intense work, we're pushing the state-of-the-art forward. Can't wait to see the amazing things the community builds with it. 🚀🚀🚀

We cooked 🧑‍🍳 a lot in improving the sounds!

We made it.