🚀 Excited to share that our paper Fast-ThinkAct has been accepted to #CVPR2026! 🎉 Efficient Vision-Language-Action reasoning via verbalizable latent planning — enabling embodied agents to think fast internally without lengthy textual reasoning. ⚡ Achieves 9.3× faster inference (89% latency reduction) than ThinkAct-7B — bringing Reasoning VLA closer to real-time robotic control. 📄 arxiv.org/abs/2601.09708 🎥 jasper0314-huang.github.io/f… 🙌 Huge congrats to @chipinhxyz, @yunzeman, @ZhidingYu, @CMHungSteven, @jankautz, Yu-Chiang Frank Wang, @FuEnYang1 #EmbodiedAI #PhysicalAI #VLA #Robotics #NVIDIAResearch @NVIDIAAI @NVIDIARobotics

It has been a blast sharing the excitement, science, art, and community of CVPR with all of you. The Publicity Chairs are officially signing off. Seattle, you’re up next! 🌲 See you at #CVPR2027! 👋 @deblinaforAI @anfurnari @CSProfKGD @YVinker @_vztu

We are presenting Fast-ThinkAct at 10:45am this morning at Poster #469. Welcome to drop by and discuss!

Train AI robots without writing a single line of code. 🤖 We just launched LeLab, the official graphical user interface for LeRobot built by @rabault_nicolas. It completely removes the command line from the robot learning workflow, taking you from raw hardware to autonomous movement visually. If you've ever wanted to get into AI robotics but were held back by complex terminal setups, this is for you. - Zero-Terminal Setup: Smart calibration with automatic USB port detection. - Easy Data Collection: Teleoperate your robot and record a dataset. - One-Click GPU Training: Don't have a massive local GPU? Scale your training instantly with Hugging Face Jobs right inside the app. Just plug in your SO-ARM101 and start teaching your robot. We put together a complete, step-by-step video guide showing exactly how to get started and train your first policy. Docs: huggingface.co/docs/lerobot/… GitHub: github.com/huggingface/leLab

Self-driving that doesn’t just drive - it reasons and explains. 🚗 🧠 ✳️=🅰️ NVIDIA announced Alpamayo 2 Super, the most powerful open reasoning VLA model for Level 4 robotaxis. 32B parameters (3x larger than Alpamayo 1 Nano), built on Cosmos. It outputs trajectories Meta-Actions Chain-of-Causation (CoC) traces, delivering interpretable explanations for its decisions. Press release: nvidianews.nvidia.com/news/n… GitHub (updating): github.com/NVlabs/alpamayo

NVIDIA announces the first open humanoid robot reference design built for robotics research. The NVIDIA Isaac GR00T Reference Humanoid Robot combines the @UnitreeRobotics H2 humanoid robot, @SharpaRobotics Wave five-fingered hands for dexterous manipulation, Jetson Thor onboard compute, and Isaac GR00T open software and models, giving researchers a full-stack platform from data capture to model deployment. Read the #NVIDIAGTC Taipei announcement: nvda.ws/4ef9VOr

Introducing Cosmos 3: Our latest frontier model for Physical AI Cosmos 3 is the world’s first fully open omnimodel with native vision reasoning, world and action generation. Today we’re releasing Super (32B) and Nano (8B) variants.

Planning your #CVPR2026 schedule for next week? 🤔 Stop by the NVIDIA booth #211 to check out demos in AV, healthcare, robotics, and vision AI. Plus, catch our research lightning talks, networking opportunities, and partner sessions across the show. 💬 Browse the full NVIDIA CVPR 2026 program here: nvda.ws/43CmXiZ See you in Denver. ✈️

Qwen-VLA feels like one of the first real robotics foundation models. A single system trained across robot manipulation, navigation, egocentric human video, simulation, and vision-language reasoning instead of isolated robot policies.

This #CVPR2026 paper from our research team is trending #1 on @HuggingFace 🤗 Meet LocateAnything: a vision-language detection model that rethinks bounding box prediction. For AI agents and robots, “seeing” is only useful if a model can pinpoint where something is fast enough to act. Trained on 138M high-quality samples, LocateAnything decodes bounding boxes in parallel instead of one coordinate at a time, improving localization accuracy while dramatically increasing throughput for visual grounding and detection. Project page: nvda.ws/4dKSohb

How can VLAs achieve 95 % reliability? Using RL post-training with EXPO-FT: - π0.5 improves to 30/30 success on all 8 tasks tested - uses only 19 min of RL data on average Paper & videos: pd-perry.github.io/expo-ft/

Thrilled to see TOPReward integrated into @LeRobotHF ! Since releasing this zero-shot approach, we've been blown away by community adoption — it's emerged as one of the top universal reward models, with applications reaching well beyond robotics. With HF's support, we can't wait to see what people build with TOPReward next!

Come to the T4V Workshop next week at CVPR for the latest developments in Vision Transformers and Multimodal AI!

93 workshops have been accepted for #ECCV2026 and you can now find a list of them all right here 📃 eccv.ecva.net/Conferences/20…

What comes after today’s visual backbones? At T4V @CVPR 2026, we’re bringing the community together for a focused half-day workshop on Transformers for Vision and Multimodal AI — covering image, video, 3D, MLLMs, efficient attention, SSMs/Mamba, and the next generation of visual architectures. 📍 Wed June 3 · Room 607 🕐 1:45–5:40 pm (Denver local time) Invited speakers: @RanjayKrishna, @thoma_gu, @sherryyangML, @jcniebles, @liuzhuang1234, and @TongPetersb. Join us at CVPR: sites.google.com/view/t4v-cv… @UNC @NVIDIAAI @NVIDIAAIDev @AIatMeta @ImagineEnpc @NJU1902 @BaskinEng #CVPR2026 #T4V #MultimodalAI #NVIDIA

Spotted: A spontaneous stop from NVIDIA CEO Jensen Huang brought crowds out to Taipei’s Raohe St. Night Market tonight.

It's not a trip to Taipei without a night market stop. 🍧📸🤳

Excited to announce this call for papers @ #ECCV2026 Workshop on Visual Perception and Reasoning in the Interactable World! Come one, come all to perception work pushing the narrative on the future of these systems: embodied, interactive, and beyond👀