NVIDIA AI Dev Weekly: Building Smarter AI Agents, AI Global Weather Forecasting, and More
Welcome to your weekly drop of developer news. From technical deep dives to events and trainings, stay up to date on the latest in generative AI, data science, accelerated computing and more.
In Case You Missed It
📣 Just Released: 26M+ Lines of Synthetic Data
Used to train our latest open reasoning model, Llama Nemotron Super v1.5, over 26M lines of synthetic data are now available to help you build your own models without expending the effort and time required to produce your own datasets.
🧠 Train Your Own Reasoning Model in a Weekend
Just published: full video tutorial teaching you how to train a small, effective reasoning model in 48 hours — all on a single GPU.
🏆 "Best Paper Honorable Mention" at ACL 2025
NVIDIA’s NeKo received "Best Paper Honorable Mention" at ACL 2025 for its task-guided MoE model. We’re elated to celebrate their multi-modal progress– advancing both state-of-the-art performance and new forms of pre-training data.
✨ Reserve Your DGX Spark
Prototype, fine-tune, and easily deploy with your own personal AI supercomputer — right on your desk. Reserve now.
Technical Deep Dives
Building More Accurate and Efficient AI Agents
The NVIDIA Nemotron family builds on the strongest open models in the ecosystem by enhancing them with greater accuracy, efficiency, and transparency using NVIDIA open synthetic datasets, advanced techniques, and tools. This enables the creation of practical, right-sized, and high-performing AI agents. Continue Reading
Fast and Accurate Large Ensemble Weather Forecasting
FourCastNet3 (FCN3) is the latest AI global weather forecasting system from NVIDIA Earth-2. FCN3 offers an unprecedented combination of probabilistic skill, computational efficiency, spectral fidelity, ensemble calibration, and stability at subseasonal timescales. Its medium-range forecasting accuracy matches that of leading machine learning models, such as GenCast, and exceeds that of traditional numerical weather prediction systems, such as IFS-ENS. Continue Reading
Bringing Verifiable Trust to AI Models
AI is entering a new era—one defined by agents that reason, plan, and take action. These agentic systems dynamically interact with APIs, tools, and even the physical environment, which introduces new complexity and vastly expands the AI attack surface and potential risks. A single compromised model can now influence downstream decisions, access external systems, trigger cascading failures, or even physically damage equipment or harm humans. Trust in model integrity can no longer be assumed. It must be verifiable. Continue Reading
Turning Complex Documents into Usable Data
Enterprises generate and store vast amounts of unstructured data. Extracting meaningful insights from this data remains a challenge for traditional optical character recognition (OCR) technologies. NVIDIA NeMo Retriever Parse overcomes the shortcomings of OCR technologies. Designed to tackle the hardest aspects of document intelligence, NeMo Retriever Parse is an optimized model built on cutting-edge vision language model (VLM) technology. Continue Reading
Semantic Prompt Injections Aren’t Just About Text Anymore.
Prompt injection, where adversaries manipulate inputs to make large language models behave in unintended ways, has long posed a threat to AI systems since the earliest days of LLM deployment. While defenders have made progress securing models against text-based attacks, the shift to multimodal and agentic AI is rapidly expanding the attack surface. This is where red teaming plays a vital role. The NVIDIA AI Red Team proactively simulates real-world attacks to uncover emerging threats in production-grade systems. Continue Reading
Developer Resources
Webinars, Trainings, and Certifications
PyTorch 2.8 Live Release Q&A: Learn about PyTorch packaging, exploring the release of wheel variant support as a new experimental feature in the 2.8 release.
AI for All: Empowering Educators and Learners: Explore how educators and learners across disciplines are leveraging NVIDIA's robust suite of resources to grow professional, strengthen academic programs, and enable cutting-edge student research.
Events
NVIDIA at SIGGRAPH 2025 | Vancouver, Canada | 10-14 August
Connect
The pace of AI advancement is accelerating rapidly, reshaping industries and pushing the boundaries of what's possible. NVIDIA continues to lead the charge!