NVIDIA Healthcare

Check out resources👇 🔸Check out Fold-CP nvda.ws/4pexJWJ 🔸NVIDIA Transformer Engine (TE) Accelerates your Transformer Architectures nvda.ws/4h6ZSNk 🔸Learn how to integrate TE into your models from BioNeMo Recipes nvda.ws/3RykDHn 🔸cuEquivariance is open-source and available nvda.ws/44SVL02 (2/2)

Congrats to @biohub on ESMFold2 - a new protein structure model trained end-to-end on 256 NVIDIA H100 Tensor Core GPUs. The full NVIDIA CUDA-X stack made it possible: ⚡ cuEquivariance fused kernels for triangle multiplication - reduced compute overhead at the core of the attention mechanism ⚡ Fold-CP context parallelism - 4,000-residue proteins on 4× H100s; 6,500 residues on 16× H100s. Longer chains. Same hardware. ⚡ NVIDIA TransformerEngine FP8 low-precision training - faster iteration, lower memory footprint, no accuracy tradeoff Longer proteins. Faster folding. (1/2)

Learn more about how IPD built it on the NVIDIA stack: ▫️cuEquivariance → geometry-aware ops at the model's core nvda.ws/3QMxQw3 ▫️BioNeMo Agent Toolkit → IPD among the first testers nvda.ws/3TuzZx5 ▫️MMseqs2-GPU → GPU-accelerated sequence search nvda.ws/4hc5c1Q

⚡ The @UWproteindesign is accelerating the next generation of protein design. Using NVIDIA cuEquivariance and TensorRT, IPD sped up RF3 structure prediction from 5.8s → 2.5s on a 256-residue protein, a >2× speedup, right at the geometric core of their model. 🧬 See how nvda.ws/4pmFOc4 (1/2)

Autonomous scientific discovery is becoming self-improving. But agents can only improve as far as the systems evaluating them. The Red Queen Gödel Machine addresses this by co-evolving agents and their evaluators - allowing the agents to continuously improve while remaining anchored to trusted ground truth. On coding tasks, it exceeded the prior baseline while using 1.35-1.72× fewer search tokens. 🧬 Bringing self-improving agents to biology and chemistry will require domain-specific tools, models and evaluation loops. That’s the future we’re building toward with BioNeMo Agent Toolkit. In paper-review experiments, pairing NVIDIA Nemotron 3 Ultra worker agents with a frontier meta-agent approached frontier-model-only performance at roughly 13× lower search-token cost. 📰 nvda.ws/3TcC7tf

Biomolecular structure prediction is a full-stack challenge, not just a model benchmark. See how NVIDIA accelerates structure prediction with GPU MSA search up to 177× faster, OpenFold3 inference up to 4× faster on NVIDIA Blackwell GPUs, and Fold-CP scaling to 32,000-token complexes across 64 NVIDIA B300 GPUs, composed into agentic workflows with BioNeMo Agent Toolkit. 🧬 Read the full blog nvda.ws/4yfAn2o

I just found a super cool date-stamped post by NVIDIA NVIDIA Healthcare from 11 years ago. Human memory is short but AI will always remember. I am very thankful to the folks at NVIDIA who really helped us in the early days. A small piece of free equipment took us very far. I am also super thankful to Phil Eisler who brought us into this game, Mark Berger who provided the samples and Kimberly Powell who recognized our work as important. We have some cool new data to share that will blow your mind in the most positive way that will help answer the question "Can NVIDIA help make people younger?". We want to be the first to produce the first conclusive proof that it does. And yes, it is over 11 years ago and I was wearing a jacket with a tie and had a weird haircut. I was 35. And at that time, if I were to invest in Nvidia (which I thought was too expensive back then), I would've made enough money for several 0-to-approval therapeutic programs that I could have funded myself. Longevity biotechnology will the be world's largest market in the future. You can already see that from the sales of Lilly and Novo - this is just the beginning. #GTC15

Don't miss out on the latest innovations - secure your spot now during early bird registration and save up to €400. Check out the details 👇

Foundation models are reshaping computational biology. Adapting models to a specific task is nontrivial, so to reduce the difficulty of building these workflows, @nvidia BioNeMo Recipes provide step-by-step training recipes built on familiar PyTorch, Hugging Face, and other patterns. This post walks through two case studies that show how the same parameter-efficient and readable recipe applies across biological modalities. Read the full post: bit.ly/44AhOst

BioNeMo Agent Toolkit, a rising star for agent-ready tools and skills for life sciences 💫 nvda.ws/4gnWehO

🌍 20 million lives impacted every single year. That is the power of technology when it is purposefully engineered to serve humanity. In many low and middle-income countries, healthcare professionals face immense barriers such as unreliable internet, a lack of localized content, and language gaps that can delay critical care. To bridge this gap, Techies Without Borders built an AI-powered experimental platform that acts as a localized "doctor's assistant." Frontline workers can now type in a medical query and receive an instant, evidence-based answer in their own language. Thanks to NVIDIA AI and accelerated computing, medical insights that used to take 5 agonizing minutes to load are now delivered in just 2 seconds. And because it runs locally on edge devices, a lack of internet connection is no longer a barrier to saving lives. This isn't just a technical achievement; it's a lifeline for clinics facing severe physician shortages.

✨Your front-row seat to our #BIO2026 special address @kpowgerade introduced BioNeMo Agent Toolkit and sat down with @sajithw (@benchling), Josh (@chaidiscovery) , and @andrewwhite01 (@EdisonSci) on how AI agents are reshaping biotech R&D and lab workflows. ▶️ nvda.ws/3Tfc4l7

🦅 American innovation is paving the way for a healthier nation. Homegrown AI companies like @AbridgeHQ and @aidocmed are transforming our healthcare system so clinicians can focus on what matters most: patient care. 📝 Abridge is tackling burnout by deploying a purpose-built foundation model across 300+ U.S. health systems, processing over 2.5 million clinical conversations weekly to eliminate after-hours paperwork. 🩻 Aidoc’s aiOS platform has analyzed over 50 million patient scans in the U.S. alone, drafting preliminary reports to help radiology teams manage rising demands. Equipping our doctors with the best tools available helps them provide a healthier future for every patient.

NVIDIA BioNeMo Agent Toolkit is now integrated with Anthropic’s Claude Science. Life sciences organizations increasingly need AI that goes beyond chat — agents that can reason over scientific questions and act on them using real computational tools. NVIDIA BioNeMo Agent Toolkit gives Claude the domain-specific scientific context to select the right tool, prepare valid inputs, execute workflows and interpret results. blogs.nvidia.com/blog/claude-sc…

(2/2) Try Litmus Bench 👇 huggingface.co/datasets/nvidi…

🧠 Most LLMs can recall chemistry facts. Few can reason about the science. ⚠️ The problem: Most chemistry benchmarks are multiple-choice, and up to 1 in 5 answers are wrong. High scores often don’t reflect true understanding. ✅ The solution: Litmus Bench fixes this with exact question-answer pairs grounded in verified chemistry data. 🔓 Now open-source with a NeMo Gym environment, Litmus Bench is the chemistry boost used by Nemotron 3 Ultra and is ready to add to your training and evaluation stack. Let's build models that think in molecules. (1/2)

The future of bio is powered by faster data Introducing the Medra AI Experimentalist: an agent that turns goals into experimental designs, learns from every result, and develops the next assay Excited to collaborate with @DARPA and @NVIDIAHealth on the future of science