See how AI agents help create a simulated obstacle course for Microduck, using NVIDIA Omniverse libraries for 3D asset creation, Newton for physics simulation, NVIDIA Isaac Lab for policy training, and Isaac Sim on NVIDIA RTX Spark for validation.

Sound on 🔊 Listen to the difference. Our new open multilingual NemotronLabs RE-USE model from NVIDIA researchers cleans up background noise, echo and other audio distortions to make speech clearer. All with just 9.6M parameters. Try the demo and download the model on @huggingface →

We've just made a breakthrough: the first true omni model running 100% locally in a browser... fully end-to-end, with no separate ASR or TTS models needed. With custom WebGPU kernels, we're able to run LFM3-Omni-2B at ~280 tok/s on my M4 Max, accepting visual, audio, or text and producing interleaved text and audio tokens. Just 90 ms to the first token and 140 ms to the first audio. The future of on-device multimodal AI agents is going to be wild.

1/4 Meet Qwenputer: Qwen3.5-4B with a full RISC-V CPU inside its weights, using the model's own Gated DeltaNet layers as RAM. It runs C and MicroPython in its forward passes, no external runtime. On DeepMind math: base 55–90%, base + Python tool 95–100%, Qwenputer 97.5–100%.

Hugging Face researcher @evijit reveals how Carbon-X, a 14B open-source AI model, rivaled GPT-6.1 Sol on genomics benchmarks for a fraction of the cost: "On many science benchmarks, closed models are ahead and open models are falling behind." "When you try to make a model better for science, you can’t really do that with existing architecture, with existing data." "For Carbon-X, we had to take the Qwen3-14B model and mid-train it with 300 billion tokens of English and DNA. RL fine-tuning will not help you. That made it comparable on genomic benchmarks with GPT 6.1 Sol, for a tiny, tiny fraction of the price." "Academics don’t have the kind of resources that these closed labs do." @huggingface

Today we're releasing the Reachy Mini Level UP update, which mainly contains three things: • A hoverboard, to give Reachy Mini some much-needed mobility (I'm a big inverted pendulum fan and I absolutely love this one, sometimes I just do sit → stand up → sit → stand up on repeat) • A smartphone app, available on iOS and Android: no computer needed to interact with the robot anymore • NFC tags, allowing for new angles of creative personalization (e.g. one hat = one personality) We hope you'll build great things with these new tools, learn stuff and have fun! Congrats to everyone involved in the release! PS: There's been a major software update that has been robust for a while now but hasn't gotten a lot of visibility yet: JavaScript apps. I encourage all our AI builders to have a look at it, because for a bunch of applications, it's so comfortable compared to Python. I think the best example of this is the Telepresence app (link below), which allows you to control your Reachy Mini and stream audio and video from your phone, even when you're not home. Also, you just click on the link and it works in your browser, no installation needed.

Reachy Mini can finally move! Today We’re launching Reachy Mini "Level Up", a set of new accessories for Reachy Mini: wheels (!), phone apps, and hats that change its personality obviously the first thing Reachy Mini did was race against Microduck (spoiler: it's faster)

Reachy Mini has wheels now! Our little open-source AI robot can finally explore beyond the desktop. And that's not all: 📱 New mobile app: no computer needed 👑 NFC hats that unlock new personalities and apps Physical AI just got a lot more fun. Watch it👇

NVIDIA is releasing an open speech model that can transcribe a meeting, label who said what, translate it and answer questions about the recording: Nemotron 3.5 Transcribe supports: - Transcription in 40 languages - Translation between English and 31 languages - Up to eight speakers, with speaker labels in 13 languages - Recordings up to an hour long You can give it names and technical terms to guide the transcript, or ask questions about the audio. Audio Q&A is trained in English. It has roughly 30B parameters, with about 4B active per token, and can run on your own NVIDIA GPU. According to NVIDIA's launch materials, it ranks first among open-weight models on Hugging Face's English Open ASR leaderboard and leads VoiceArena's 14-system test of transcription with speaker attribution. Built for recorded audio. Live streaming uses a separate model. Super cool to see! Model: Demo: Blog:

Introducing Samaya Sterling, our new model post-trained for challenging financial research that outperforms the frontier! Sterling started off as a two person research effort to see if we could increase efficiency in agentic orchestration. To our surprise we were able to post-train a much smaller model (Nemotron 3.5 Lightning, a 30B mixture-of-experts model with 3B active parameters) to greater accuracy (58.9% vs 55.9% on FrontierFinance) and 11x lower cost ($0.61 vs $6.89) than Claude Fable 5.1, the best performing frontier model. We did this via a four-stage training process, two stages of SFT where the model first learns to research with our financial tools and how to construct a good agentic research strategy. The second two stages are RL, with an offline GRPO stage where the model further hones its research strategy from eight teachers that provide diverse approaches, and an online GRPO stage where the model can make end to end attempts and learn from its mistakes. We see substantial gains from the co-adaptation of the model and harness through post-training, both through the model learning to effectively use the high-quality financial tools, and better compaction approaches that let even a small model execute on longer and deeper research. We are exploring multiple follow-up directions, from multi-agent research swarms using this small model, to training firm-specific models to encode unique judgement, to training a larger model, where early results show even greater gains from the training process. The full research is linked below!

Meet NVIDIA Nemotron 3.5 Transcribe 🎙️ An open-weight, promptable 30B MoE model with ~4B active parameters that brings transcription in 40 languages, speaker attribution, translation, contextual prompting and audio understanding into one checkpoint. Watch it keep track of who said what when everyone’s talking at once. 1/ 🧵

A robot duck walking through simulated snow! 🦆 MicroDuck from @huggingface walking through a snowy scenery simulated in NVIDIA Newton with a custom snow solver. We all know that this robot duck is quite comfortable on the stage, but right now we can see it's quite capable in the snow too! ❄ The snow is done properly, and it's very important to go more in details here. Flake sizes, masses and fall speeds come from measured snowfall literature going back to the 1970s. Flakes tumble like thin plates. The rate is a real 8 mm/h liquid equivalent, around 290,000 flakes airborne at once. Snow sticks according to the angle of repose at air temperature, rides along in each link’s own frame as the robot moves, and the feet press footprints into a 2 mm-resolution snowpack. The walking policy was trained in Isaac Lab and reimplemented in pure Newton, without the Isaac Lab runtime. Then checked against Isaac Lab running headless: observations matching to 1e-6, actions to 2e-6, an identical servo model across 4,088 random inputs, walking speed within 0.003 m/s. Twenty-two of twenty-two checks! ✅ Rendering snow is achievable!!! @TheRobotSimGuy 🐐, @NVIDIARobotics ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →

🤖🪿 Microduck looks like a toy but runs real AI walking policies you can retrain yourself The $399 robot from Hugging Face’s team has 15 motors, onboard sensors, and simulation-trained moves. Perfect little walker for anyone who wants a desk robot that might already be owned by a rogue swarm of ChatGPT agents 😂 Writer: Lucas

📦 NVIDIA did it first, now we are releasing 2,600+ free game-ready 3D models (+200 each week). For good. 🎛️ Each model is a program, with parameters you can tweak to customize look, state, season, color, wear, etc. All thanks to the $500,000 in funding we received from @xprize and @GeminiApp and all of our early individual funders, thank you all! Go build.

Carbon-X is the first model achieving frontier intelligence on DNA understanding! It is fluent in both English and DNA so you can chat to DNA! Biology and DNA understanding will have a huge impact and be the main focus of the frontier labs. Model: