---
title: "NVIDIA Nemotron 3.5 Lightning Delivers Fast, Accurate Specialized Task Execution for Long-Running Agents"
description: "NVIDIA Nemotron 3.5 Lightning is an open 30B mixture-of-experts model with 3B active parameters, designed for the high-volume execution layer of long-running autonomous agents. The post positions it alongside frontier reasoning models: larger systems handle planning and orchestration, while Lightning handles tool calls, result validation, subagent delegation, and other routine work. Its design combines harness-optimized training, multi-token prediction for speculative decoding, DFlash and DSpark draft models, and NVFP4 quantization alongside BF16. The model is customizable through LoRA or full SFT, reinforcement learning, and environment-based evaluations, with weights, training data, and recipes released under OpenMDW-1.1. The post reports up to 4x the output speed of similar-sized models, 86% accuracy on PinchBench, and completion of 10,000 tasks 30% faster than Qwen3.6 35B at similar accuracy, while NeMo Switchyard routes work among models."
---

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

[NVIDIA](https://yomu.fyi/company/nvidia-developer-blog) · Tanya Lenz · Aug 11, 2026

**Type:** Announcement

## Summary

NVIDIA Nemotron 3.5 Lightning is an open 30B mixture-of-experts model with 3B active parameters, designed for the high-volume execution layer of long-running autonomous agents. The post positions it alongside frontier reasoning models: larger systems handle planning and orchestration, while Lightning handles tool calls, result validation, subagent delegation, and other routine work. Its design combines harness-optimized training, multi-token prediction for speculative decoding, DFlash and DSpark draft models, and NVFP4 quantization alongside BF16. The model is customizable through LoRA or full SFT, reinforcement learning, and environment-based evaluations, with weights, training data, and recipes released under OpenMDW-1.1. The post reports up to 4x the output speed of similar-sized models, 86% accuracy on PinchBench, and completion of 10,000 tasks 30% faster than Qwen3.6 35B at similar accuracy, while NeMo Switchyard routes work among models.

## Context

Long-running AI agents spend much of their time on high-volume execution tasks such as tool calls, result validation, and subagent delegation. Using a frontier reasoning model for every step adds cost and latency, motivating a division between complex planning and more efficient execution.

## Approach / What changed

Nemotron 3.5 Lightning uses a 30B mixture-of-experts architecture with 3B active parameters, harness-optimized training, multi-token prediction, speculative-decoding draft models named DFlash and DSpark, and NVFP4 quantization alongside BF16. The release also provides weights, training data, and recipes for customization, while NeMo Switchyard routes tasks to Lightning or other models according to the work required.

## Takeaways

- The 30B mixture-of-experts model activates 3B parameters per token, with a router sending each token to only a few experts.
- Multi-token prediction is included during training and followed by an MTP-boosting phase; DSpark and DFlash are provided as draft models for speculative decoding.
- On PinchBench, the post reports 86% accuracy and completion of 10,000 tasks 30% faster than Qwen3.6 35B at similar accuracy.

**Tags:** [Deployment](https://yomu.fyi/topic/deployment), [LLMs](https://yomu.fyi/topic/llm), [Machine Learning](https://yomu.fyi/topic/machine-learning), [Open Source](https://yomu.fyi/topic/open-source), [Performance](https://yomu.fyi/topic/performance)

- Source: [NVIDIA](https://developer.nvidia.com/blog/nvidia-nemotron-3-5-lightning-delivers-fast-accurate-specialized-task-execution-for-long-running-agents)
- Source URL: https://developer.nvidia.com/blog/nvidia-nemotron-3-5-lightning-delivers-fast-accurate-specialized-task-execution-for-long-running-agents
- Ingested by Yomu: 2026-08-28T08:01:01.574Z

[Read original post](https://developer.nvidia.com/blog/nvidia-nemotron-3-5-lightning-delivers-fast-accurate-specialized-task-execution-for-long-running-agents)