Check out the latest FastGen release, now with support for a ton of new image and video models!
This release also includes AYF training code and checkpoints for FLUX.1 [dev] and Qwen-Image. Finally! 🎉
Tired of waiting for your diffusion models?
🚢 We just shipped a major FastGen update:
- new models: H3, LTX-2.5, LingBot-World, Cosmos 3, FLUX.2
- new methods: PDD, AYF, AnyFlow, Causal Forcing
- new features: multimodal distillation, image editing, camera conditioning, KV cache management, ComfyUI
🔗 Try it out here: github.com/NVlabs/fastgen
Amirmojtaba Sabour retweeted
Avengers assemble: 1 year BakerLab + NVIDIA GenAIR = 🔥 rfd4-proteina.org
Presenting RFD4-Proteina, our latest protein design model, significantly better and faster than everything we compared against, both in silico and in the lab! 1/30
Amirmojtaba Sabour retweeted
We’re hiring PhD interns for NVIDIA Research's Learning and Perception Research (LPR) team!
We are passionate about embodied AI research that pushes boundaries while making real-world impact. We are particularly interested in agentic reasoning, multimodal and continual learning, AI scientists, and beyond.
To be considered, you must be a currently enrolled PhD student with a promising research track record (publications and/or core contributor to open-source projects). Please apply directly at the link below!
Listing: nvidia.wd5.myworkdayjobs.com…
Team homepage: research.nvidia.com/labs/lpr…
Amirmojtaba Sabour retweeted
Please help spread the word!
We’re hiring Research Scientists and Research Interns at NVIDIA Research to work on some of the most exciting problems in generative AI:
🌎 World Models & Video Diffusion
🧠 Diffusion Language Models
🧬 Biomolecule Design
If your research aligns with any of these areas, please apply directly through the links below.
Open positions + application links:
research.nvidia.com/labs/gen…
Amirmojtaba Sabour retweeted
"Intelligence from Learnable Novelty "
This paper suggests that intelligence is not surprise itself, but the ability to turn surprise into understanding.
Since novelty search gets trapped by noise, and surprise minimization gets trapped by stasis, it splits surprise into learnable and unlearnable parts, then optimizes only the learnable part.
With a cheap differentiable estimator, it finds rule 110, grows solitons, clusters MNIST without labels, and improves RL exploration.
Amirmojtaba Sabour retweeted
The latent-vs-pixel debate misses the point.
GPT Image 2 shows what users notice: pixel-level fidelity.
Latent models show what scales: compact semantic structure.
We connect them by replacing VAE/RAE decoders with a Pixel Diffusion Decoder.
Code and Model available: research.nvidia.com/labs/sil…
🧵(1/N)
Amirmojtaba Sabour retweeted
🚀 🚀 🚀 Excited to share our new paper:
Remember to be Curious: Episodic Context and Persistent Worlds for 3D Exploration
What does it take for an agent to stay curious in a 3D world?
The answer is memory.
🌐 Project: recuriosity.github.io/
📄 Paper: arxiv.org/abs/2605.22814
💻 Code: github.com/recuriosity/recur…
Amirmojtaba Sabour retweeted
My first blog post in over a year is a deep dive on flow maps🗺️, or how to learn the integral of a diffusion model to enable faster sampling and several other cool tricks.
It's the longest one yet👀 Let me know what you think!
sander.ai/2026/05/06/flow-ma…
Amirmojtaba Sabour retweeted
🎆⭐️You thought 2025 would end quietly? Think again. One last exciting update: Tilt Matching!⭐️🎆
We propose a simple, scalable algorithm to sample unnormalized densities and fine-tune generative models
📄: arxiv.org/abs/2512.21829
With @brianlee_lck @msalbergo
(1/5)
Amirmojtaba Sabour retweeted
Check out our illustrations, visualizations and applications of Flow Map Trajectory Tilting to complex problems on the project page! 🔥
flow-map-trajectory-tilting.….
This was an amazing collaboration with @amsabour, @msalbergo, @cdomingoenrich, @nmboffi, @FidlerSanja, @evdende2 ! 😊
Excited to share our latest work on test-time scaling for diffusion models. We propose FMTT, a flow-map-based look-ahead that enables efficient reward-guided sampling along the diffusion trajectory. Check out the paper for details.
⭐️New Paper⭐️ just in time for NeurIPS. Presenting: Test-time scaling of diffusions with flow maps.
project: flow-map-trajectory-tilting.…
arXiv: arxiv.org/abs/2511.22688
Want to properly adapt your diffusion at inference time with a reward? You should do everything at the level of the flow map.
With the amazing @amsabour, @cdomingoenrich, @nmboffi, @FidlerSanja , @karsten_kreis, and @evdende2.
Amirmojtaba Sabour retweeted
⭐️New Paper⭐️ just in time for NeurIPS. Presenting: Test-time scaling of diffusions with flow maps.
project: flow-map-trajectory-tilting.…
arXiv: arxiv.org/abs/2511.22688
Want to properly adapt your diffusion at inference time with a reward? You should do everything at the level of the flow map.
With the amazing @amsabour, @cdomingoenrich, @nmboffi, @FidlerSanja , @karsten_kreis, and @evdende2.
Amirmojtaba Sabour retweeted
🕹️We are excited to introduce "ChronoEdit: Towards Temporal Reasoning for Image Editing and World Simulation"
ChronoEdit reframes image editing as a video generation task to encourage temporal consistency. It leverages a temporal reasoning stage that denoises with “video reasoning tokens” to "reason" on physically plausible edits.
See the attached video for results.
Project Page: research.nvidia.com/labs/tor…
Arxiv: arxiv.org/abs/2510.04290
Code and model are coming.
Amirmojtaba Sabour retweeted
Super excited to share what @stephenz_y and I’ve been up to during our internship at🍎:
Using optimal transport makes flows straighter and generation faster in flow matching, but small batch OT is biased and large batch OT is slow.
What to do? Use semidiscrete OT!
🧵
Excited to share our latest work on diffusion distillation. We propose AYF, a flow map framework that generalizes consistency models, flow matching, MeanFlow models, and more. Check out the project page for details.
📢📢 "Align Your Flow: Scaling Continuous-Time Flow Map Distillation"
New flow map framework for state-of-the-art few-step generation, w/ the amazing @amsabour and @FidlerSanja.
🔥 Project page: research.nvidia.com/labs/tor…
📜 Paper: arxiv.org/abs/2506.14603
🧵Thread below...
(1/n)
Amirmojtaba Sabour retweeted
📢📢 "Align Your Flow: Scaling Continuous-Time Flow Map Distillation"
New flow map framework for state-of-the-art few-step generation, w/ the amazing @amsabour and @FidlerSanja.
🔥 Project page: research.nvidia.com/labs/tor…
📜 Paper: arxiv.org/abs/2506.14603
🧵Thread below...
(1/n)
Amirmojtaba Sabour retweeted
Nvidia just announced Align Your Flow
Scaling Continuous-Time Flow Map Distillation
Amirmojtaba Sabour retweeted
We are excited to share Cosmos-Drive-Dreams 🚀
A bold new synthetic data generation (SDG) pipeline powered by world foundation models—designed to synthesize rich, challenging driving scenarios at scale.
Models, Code, Dataset, Tookit are released.
Website: research.nvidia.com/labs/tor…
Amirmojtaba Sabour retweeted
FlexiCubes is now under Apache 2.0! 🎉
We've been excited to see FlexiCubes extracting high-quality meshes across the community in projects like TRELLIS and TripoSF --- now it's available with a more permissive license.
Let's keep building. 💙
👉 Flexicubes is in NVIDIA Kaolin: github.com/NVIDIAGameWorks/k…
👉 Or a standalone repo github.com/nv-tlabs/FlexiCub…

























