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

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.

Takeaways

1. 1

   The 30B mixture-of-experts model activates 3B parameters per token, with a router sending each token to only a few experts.

2. 2

   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.

3. 3

   On PinchBench, the post reports 86% accuracy and completion of 10,000 tasks 30% faster than Qwen3.6 35B at similar accuracy.