BF16 vs NVFP4 with Nemotron 3.5 Lightning @NVIDIADeveloper
BF16 vs NVFP4 with Nemotron 3.5 Lightning  @NVIDIADeveloper
Uploaded September 2026 | Updated November 2026, 3 days ago
Learn how NVFP4 compares to BF16 in practice, why the smaller checkpoint frees up GPU memory without a big quality hit, and where each format still makes the most sense.

Using Nemotron 3.5 Lightning, which ships as both a BF16 and an NVFP4 checkpoint, we look at memory footprint, throughput, and accuracy, then walk through a notebook that runs the whole comparison on a single H100 so you can reproduce these results.

What Is NVFP4?: youtu.be/UTfg-_EGurw?si=8V9kG1AJi4O4NIFB

Notebook: github.com/patelmiteshn/inference-cookbooks/blob/main/examples/nvfp4_vs_bf16.ipynb
BF16 vs NVFP4 with Nemotron 3.5 LightningHow to Connect Two DGX Sparks with NVIDIA SyncAsk the Experts: Nemotron 3.5 Super VL | Nemotron LabsAsk the Experts: Nemotron 3 Nano Omni | Nemotron LabsHow We Built Self-Evolving Hermes Agents With NVIDIA NemoClawLong-Running AI Agents: The Next Breakthrough in Enterprise WorkDense vs. MoE: How to Choose the Right AI ArchitectureHow to Post-Train NVIDIA Cosmos 3 for Robot Action PredictionA New Way to Build Edge AI: Agentic Development on NVIDIA JetsonHow the Developer Community Builds Sub-Agents with NVIDIA Nemotron 3 Nano Omni | Nemotron LabsBuild a Visual AI Agent in Under 30 Minutes with NVIDIA VSS Blueprint 3.3Build Visual AI Agents From a Prompt With NVIDIA Cosmos and VSS 3.3
NVIDIA Developer |

BF16 vs NVFP4 with Nemotron 3.5 Lightning

SHARE TO X SHARE TO REDDIT SHARE TO FACEBOOK WALLPAPER