NVIDIA Nemotron 3.5 Lightning Delivers Fast, Accurate Specialized Task Execution for Long-Running Agents

NVIDIA released Nemotron 3.5 Lightning, an open 30B mixture-of-experts model with 3B active parameters, optimized for high-volume, low-latency execution in long-running AI agents. The model, part of the Nemotron 3 family, includes speculative decoding and harness-optimized training, and is supported by the NVIDIA NemoClaw open source stack. NVIDIA also introduced NeMo Switchyard, a library for routing tasks to the best model.

Long-running AI agents https://www.nvidia.com/en-us/glossary/ai-agents/ spend most of their time on high-volume execution: tool calls, result validation, and subagent delegation. Using a frontier reasoning model for every execution step adds cost and latency. NVIDIA Nemotron 3.5 Lightning https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-NVFP4 is an open 30B mixture-of-experts MoE https://www.nvidia.com/en-us/glossary/mixture-of-experts/ model with 3B active parameters built for that execution layer of always-on agents. It is designed for harnesses like OpenClaw and Hermes Agent—all supported by the NVIDIA NemoClaw https://github.com/NVIDIA/NemoClaw open source security and management stack for running always-on AI agents. The NVIDIA Nemotron https://developer.nvidia.com/topics/ai/nemotron open model family is like a software library, with each release continuously improving accuracy and speed. As these models evolve, the rapid maturation of model routing and orchestration is also underway. This is important because developers increasingly build applications with a system of models https://www.nvidia.com/en-us/glossary/multi-agent-systems/ :~:text=request%20is%20inputted.-,System%20of%20Models,-Depending%20on%20the . Frontier reasoning models https://www.nvidia.com/en-us/glossary/frontier-models/ such as Nemotron 3 Ultra https://developer.nvidia.com/blog/nvidia-nemotron-3-ultra-powers-faster-more-efficient-reasoning-for-long-running-agents/ handle orchestration and complex planning while smaller, more efficient models handle the high-volume execution layer. This post introduces NVIDIA Nemotron 3.5 Lightning and explains how its smaller MoE design is optimized for high-volume, low-latency execution in autonomous agents. It also details the inference and training innovations that power it. Finally, the post also introduces NVIDIA NeMo Switchyard https://github.com/NVIDIA-NeMo/Switchyard , a library that intelligently routes each task to the best model for the job. Why is Nemotron 3.5 Lightning ideal for long-running AI agents? Nemotron 3.5 Lightning is a customizable open 30B MoE model with 3B active parameters, providing optimal high-volume execution for autonomous agents. MoEs are fast and efficient because a router sends each token to just a few of its many experts, so only a fraction of the model’s parameters run per token. This provides the capacity of a larger dense model at the compute cost of a small one. Nemotron 3.5 Lightning is the smallest member of the Nemotron 3 model family and ships with many of the same techniques proven across the family, including: Speculative decoding : Multi-token prediction was included during Nemotron 3.5 Lightning training as for Nemotron 3 Super and Nemotron 3 Ultra . Nemotron 3.5 Lightning also ships with DFlash and DSpark, enabling more comprehensive inference optimization https://developer.nvidia.com/blog/mastering-llm-techniques-inference-optimization/ across a range of serving scenarios. Harness-optimized training: The model is trained for popular agent harnesses, enabling agents to make more accurate calls while reducing latency for high-volume tasks. The result is a model built for execution-focused, high call volumes, and low latency—all at a size that deploys anywhere from an NVIDIA DGX Spark https://www.nvidia.com/en-us/products/workstations/dgx-spark/ to a data center. Customize Nemotron 3.5 Lightning out of the box Models earn their place in specialized AI https://www.nvidia.com/en-us/glossary/specialized-ai/ agent systems when they’re adapted to the job. And Lightning-class models are highly customizable: small models fine-tune faster, cheaper, and on far more modest hardware than their larger counterparts. You can customize Nemotron 3.5 Lightning out of the box to fit your workload. As with every Nemotron open model https://www.nvidia.com/en-us/glossary/open-models/ launch, the weights, training data, and recipes are released as permissively as possible under OpenMDW-1.1, so you can: - Fine-tune with LoRA or full SFT using NeMo Automodel https://github.com/nvidia-nemo/automodel and NeMo Megatron Bridge https://github.com/NVIDIA-NeMo/Nemotron/tree/main/usage-cookbook/Nemotron-3.5-Lightning/lora-text2sql - Run reinforcement learning and environment-based evaluations and rollouts with NeMo RL https://github.com/nvidia-nemo/rl and NeMo Gym https://github.com/NVIDIA-NeMo/gym This release includes Nemotron-RL Agentic Terminal Pivot https://huggingface.co/datasets/nvidia/Nemotron-RL-Agentic-Terminal-Pivot-v1-nano35-release , an open agentic reinforcement learning dataset used to train some of the coding agent capabilities. Route work to the right model using NeMo Switchyard While frontier models may win the headlines, models like Nemotron 3.5 Lightning earn their medals in the trenches. They handle requests like git pull, validate tool outputs, format results, and run the routine calls that dominate any long-running agent’s token budget. Model routing and orchestration help make this division of labor more accessible. They are now available through NVIDIA NeMo Switchyard https://developer.nvidia.com/blog/route-ai-agent-workloads-across-models-with-nvidia-nemo-switchyard . Switchyard can expose Nemotron 3.5 Lightning as a routing target alongside your open and closed models, so every request lands on the most capable and efficient model that can handle it. Plans route up to the frontier, execution routes down to Lightning, ensuring that your tokens are spent efficiently and effectively. How does Nemotron 3.5 Lightning perform on the accuracy-speed Pareto frontier? Nemotron 3.5 Lightning delivers leading accuracy at the highest output speed in its class, winning the accuracy-versus-speed Pareto frontier on the Artificial Analysis Intelligence Index. This index combines nine evaluations to measure model performance across agentic tasks, coding, scientific reasoning, and general intelligence. Nemotron 3.5 Lightning combines strong intelligence with up to 4x output speed of similar-sized models, placing it on the accuracy-speed Pareto frontier for high-volume agent workloads. Agent efficiency ultimately comes down to how quickly a model completes useful work and not simply how fast it generates tokens. On PinchBench, Nemotron 3.5 Lightning reaches 86% accuracy while completing 10,000 tasks 30% faster than Qwen3.6 35B at similar accuracy. Higher inference https://www.nvidia.com/en-us/glossary/ai-inference/ throughput and token https://blogs.nvidia.com/blog/ai-tokens-explained/ efficiency places Nemotron 3.5 Lightning on the efficiency frontier, helping always-on agents finish high-volume work faster. How does Nemotron 3.5 Lightning deliver speed without compromising accuracy? Nemotron 3.5 Lightning delivers speed and customization without compromising accuracy through speculative decoding, and quantization. Speculative decoding Nemotron 3.5 Lightning is built to quickly generate tokens and has the ability to generate multiple tokens through speculative decoding. This is a process whereby the model, or draft model, will draft some number of tokens which are efficiently reviewed. Nemotron 3.5 Lightning underwent a dedicated pretraining stage to bake multi-token prediction MTP into the model, as with Nemotron 3 Super and Ultra. After training, a dedicated MTP-boosting phase further improved MTP accuracy. Beyond MTP, two draft models are provided with Nemotron 3.5 Lightning: DSpark https://arxiv.org/pdf/2607.05147 , which is recommended for DGX Spark inference workloads and low concurrency data center workloads. MTP is best suited for medium to high concurrency, with the optimal draft length decreasing as concurrency increases. NVIDIA is also releasing a DFlash https://developer.nvidia.com/blog/boost-inference-performance-up-to-15x-on-nvidia-blackwell-using-dflash-speculative-decoding/ draft model, which can be measured against the others and may perform best for your workloads. Quantization Nemotron 3.5 Lightning ships with an NVFP4 https://developer.nvidia.com/blog/introducing-nvfp4-for-efficient-and-accurate-low-precision-inference/ checkpoint alongside BF16, using the same specialized NVFP4 kernels that power Nemotron 3 Ultra across NVIDIA Blackwell, NVIDIA Hopper, and NVIDIA Ampere GPUs. The same file serves just as well in data centers as it does on your desktop DGX Spark. How is Nemotron 3.5 Lightning ideal for local AI? Nemotron 3.5 Lightning makes capable agentic AI accessible on local systems including NVIDIA Jetson, GeForce RTX 5090, and DGX Spark. NVIDIA has worked with a number of teams including EXO Labs to understand how this model performs on DGX Spark. In addition, you can run Nemotron 3.5 Lightning across the industry standard set of tools, including LM Studio, llama.cpp, Ollama, and Unsloth. Partner ecosystem Nemotron 3.5 Lightning is supported by a growing ecosystem of partners across harnesses, customization, deployment, and inference, including: Post-training: AgileRL https://www.agilerl.com/blog/partnering-with-nvidia-for-post-training-nemotron-models-to-beat-frontier-apis , Aible, Applied Compute, Deep Cogito, distil labs https://www.distillabs.ai/blog/the-best-mid-size-model-to-fine-tune-was-the-third-best-model-to-prompt/ , Fastino Labs http://fastino.ai/blog/fastino-nemotron-3-5-lightning-finance-and-healthcare , Local AI Labs, Prime Intellect, Reasonable, Thinking Machines Lab, Thoughtworks https://www.thoughtworks.com/insights/blog/generative-ai/putting-nvidia-nemotron-3-5-lightning-test , Trajectory http://trajectory.ai/field-notes/continual-learning-cheap-enough-to-meter , Uniphore http://uniphore.com/blog/testing-nvidia-nemotron-3-5-lightning-against-real-enterprise-agentic-workloads/ Inference software: Ollama https://ollama.com/blog/nemotron-3-5-lightning , Exo, Canonical https://ubuntu.com/blog/nvidia-nemotron-3-5-lightning , LM Studio https://lmstudio.ai/models/nvidia/nemotron-3.5-lightning , Unsloth Harnesses and agent frameworks: Cline , Factory AI https://factory.ai/news/nvidia-dgx-spark , Hermes Agent, Kilo Code https://blog.kilo.ai/nvidia-nemotron-3-5-lightning , LangChain http://www.langchain.com/blog/switchyard-agent-routing-benchmark , LM Studio Bionic, OpenClaw, OpenCode http://opencode.ai/ , OpenHands https://www.openhands.dev/blog/nvidia-nemotron-3.5-lightning , Pi Cloud service provider platforms : Google Cloud Gemini Enterprise Agent Platform https://docs.cloud.google.com/gemini-enterprise-agent-platform Model Garden , MSFT Foundry MaaP & Managed Compute , OCI Enterprise AI GSI: Accenture Japan, TCS, Tech Mahindra, Wipro AI natives: Arcos Labs https://www.arcoslabs.com/blog/arcos-law-on-premise-ai-nvidia-nemotron/ , CodeRabbit https://www.coderabbit.ai/blog/teaching-nvidia-nemotron-3-5-lightning-to-route-code-reviews , Dream, Harvey Hosted inference service providers: Baseten https://www.baseten.co/blog/introducing-nemotron-35-lightning/ , BlackBox AI, CoreWeave https://wandb.ai/inference/coreweave/cw nvidia Nemotron-3.5-Lightning-30B-A3B , Crusoe, DeepInfra https://deepinfra.com/blog/nvidia-nemotron-3.5-lightning-release , Fireworks AI http://fireworks.ai/models/fireworks/nemotron-lightning-3p5-30b-a3b , FriendliAI https://friendli.ai/blog/nvidia-nemotron-3.5-Lightning , GMI Cloud https://www.gmicloud.ai/en/blog/nvidia-nemotron-3-5-lightning-is-live-on-gmi-cloud-what-your-agent-system-was-missing , Modal, Nebius https://tokenfactory.nebius.com/endpoints?modals=endpoint-details&model-id=nvidia/Nemotron-3 5-Lightning , Together AI https://www.together.ai/models/nemotron-3-5-lightning Get start building with Nemotron 3.5 Lightning Nemotron 3.5 Lightning is fully open—weights, data, and recipes—so you can adapt it to your workflows and deploy it anywhere.To get started, try it on build.nvidia.com http://build.nvidia.com or through OpenRouter https://openrouter.ai/nvidia/nemotron-3.5-lightning:free . Download the weights from Hugging Face https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-NVFP4 , and ModelScope https://modelscope.ai/collections/nv-community/Nemotron-35-Lightning . Want to dive deeper? - Read the Nemotron 3.5 Lightning cookbook https://github.com/NVIDIA-NeMo/Nemotron/tree/main/usage-cookbook/Nemotron-3.5-Lightning - Deploy with the vLLM https://github.com/NVIDIA-NeMo/Nemotron/blob/main/usage-cookbook/Nemotron-3.5-Lightning/vllm cookbook.ipynb , SGLang https://github.com/NVIDIA-NeMo/Nemotron/blob/main/usage-cookbook/Nemotron-3.5-Lightning/sglang cookbook.ipynb , and TensorRT-LLM https://github.com/NVIDIA-NeMo/Nemotron/blob/main/usage-cookbook/Nemotron-3.5-Lightning/trtllm cookbook.ipynb deployment guides - Set up routing with Switchyard docs https://github.com/NVIDIA-NeMo/Switchyard/blob/main/docs/getting started.md Stay up to date on NVIDIA Nemotron by subscribing to NVIDIA news and following NVIDIA AI on LinkedIn , X , Discord , and YouTube . Visit the Nemotron developer page for resources to get started. Explore open Nemotron models and datasets on Hugging Face , ModelScope , and Blueprints on build.nvidia.com . Engage with Nemotron livestreams , tutorials , and the developer community on the NVIDIA forum and Discord .