A multi-year research project run across several countries found that smaller COVID-19 vaccine booster doses can produce strong immune responses that persist over time.
Clinical trials funded by @CEPIvaccines involved over 2000 people in Australia, Indonesia and Mongolia. Participants received a fraction of a standard dose of a licensed COVID-19 vaccine, like those produced by Pfizer-BioNTech and AstraZeneca.
The results showed antibody and cellular immune responses comparable to results from full-dose boosters for up to two years.
“If COVID cases were to ramp up again and there be strain on the vaccine supply, these findings could be key to expanding vaccination coverage quickly and preventing future infections and deaths. The findings could also prove helpful for future vaccines against other coronaviruses manufactured using similar technologies.” -- Dr Amol Chaudhari, Clinical Development Science Lead at CEPI.
The full story: https://t.co/G8CoaTwweQ
A hosted RL run took Nemotron 3 Nano from 22% to 91% accuracy on a math task for under $5.
Using @PrimeIntellect Lab, the whole loop runs on hosted infrastructure. You check a baseline, train the model until the reward climbs, and retest to confirm it actually learned, ending with a downloadable LoRA adapter.
You can also scale this workflow to Nemotron 3 Super and Ultra by changing one line.
We cut DeepSeek-V4 Pro startup from 8 minutes to under 2 minutes by moving weights over the fastest path to GPU memory with GPU-to-GPU RDMA.
This was achieved using NVIDIA ModelExpress (MX), the weight distribution and cache management service in NVIDIA Dynamo, and this same approach speeds up both inference and RL post-training too. MX reuses kernel caches, while inference workers fetch updated weights directly from other GPUs over NIXL—avoiding centralized broadcasts and keeping weight movement off the critical path.
We tested Nemotron 3 Ultra on agentic chip design.
The task is RTL coding: a model iteratively writes the code that defines a chip's logic, runs it through a simulator, reads the failures, and rewrites.
Across nine categories of real design work, Nemotron 3 Ultra averaged a 97.1% pass rate at 6,629 tokens per iteration, best of the open models tested on both.
We're contributing open models, weights, data and research to the Open Secure AI Alliance.
This includes NVIDIA Labs Object-Oriented Agents (NOOA): open agent harness research in the form of an object-oriented framework that reaches state-of-the-art results across software engineering, cybersecurity and reasoning benchmarks.
Developers can use the findings to optimize their own harnesses and then contribute back.
10 million downloads and counting for NVIDIA Cosmos models on @huggingface. 🤗
This milestone belongs to the developers and researchers using open world foundation models to build robots, autonomous vehicles and other physical AI systems.
Thank you for downloading, experimenting, and building with us. 💚