AI Briefing#nemotron
The latest AI and developer news about #nemotron, with the original source and a short summary.
The latest AI and developer news about #nemotron, with the original source and a short summary.
The partnership includes a sovereign deployment for Poland's e-Health platform and plans to add 15MW of NVIDIA Vera Rubin capacity.
The framework compresses Nemotron 3.5 Super from 120B to 74B parameters while retaining 95% of its mean benchmark score.
AIM Intelligence combines its Starfort platform with NVIDIA OpenShell to secure enterprise workflows, while LG AI Research opens API access for K-EXAONE-2.5 built on Nemotron data.
Reducing the 30B model size to 10GB and recovering 89.4% of benchmark scores makes it possible to run on a 24GB MacBook.
Alexiuk argues that full access to weights, data, and training recipes is essential for AI democratization, using NVIDIA Nemotron as the primary example for developers.
NVIDIA announced that Nemotron 3.5 Lightning generation speed on DGX Spark is up to 1.73 times faster when DSpark is applied.
The model features a hybrid mixture-of-experts architecture activating approximately 4.3B parameters and supports 40 languages.
The model supports input sampling rates from 8 to 48 kHz and is designed to restore speech quality while preserving fidelity across diverse degradations.
Shared computing infrastructure allows municipalities to deploy vision agents and digital twins, with ITMAX targeting 17,000+ cameras by 2027 and Linker Vision cutting design iteration time from two weeks to under an hour.
NVIDIA has released Nemotron 3.5 Transcribe, which supports 40 languages, as open source.
The 4GB model targets voice assistants and robotics on RTX, Jetson, and DGX Spark hardware.
The laptops start at $2,599 and feature a Blackwell RTX GPU and Grace CPU superchip with up to 1 petaflop of FP4 AI performance.
Open-weight model supports 40 languages and speaker diarization, with a 30B parameter architecture using 4B active parameters per token.
The 1.58-bit MoE model achieves 172 tok/s decode and 7000+ tok/s prefill speeds in both Stock llama.cpp and the PrismML fork.
NVIDIA has released Nemotron 3.5 Bolt E4B, a small model for interactive agents that operates on 4GB VRAM.
NVIDIA releases five Nemotron 3.5 models for embedding, reranking, and parsing, with the 8B Embed model topping ViDoRe v3.1 and the suite reducing agent token costs by up to 14%.
NVIDIA expands its open source Agent Toolkit with the Agent Knowledge Foundation, CUDA-X libraries, and new Kumo open models to help enterprises ground agents in company data without moving it or changing governance.
The portfolio includes Nemotron 3.5, Kumo for structured data, Cosmos 3.5 for robotics, and Ising for quantum computing, enabling customization with local data.
The Apache 2.0 licensed tool uses custom Metal kernels and speculative decoding to outperform standard MLX engines on local coding models.
The 4B parameter model uses a looped architecture with interleaved sliding-window and full attention to scale intelligence on a small footprint.
The dataset covers 22 Indic languages with native and romanized scripts, generated via an open-source NeMo Data Designer pipeline.
The model supports multimodal retrieval across 34 languages for text and 18 for images, with open commercial use rights.
NVIDIA released Nemotron 3.5 Super VL, a 120-billion-parameter open model, alongside two edge-optimized Bolt models for local agentic workloads.
The region plans to deploy 1 gigawatts of AI cloud capacity by the end of 2027.