NVIDIA-Nemotron-3.5-Lightning-30B-A3B-Base-BF16

Looking for the post-trained model? See NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16 for the full-precision release, or NVIDIA-Nemotron-3.5-Lightning-30B-A3B-NVFP4 for optimized inference.

Model Overview

Model Developer: NVIDIA Corporation

Model Dates: December 2025 - May 2026

Data Freshness:

• The pre-training data has a cutoff date of September 2025.

Description

NVIDIA-Nemotron-3.5-Lightning-30B-A3B-Base-BF16 is a large language model (LLM) trained by NVIDIA. This is the base (pre-trained) checkpoint of the Nemotron 3.5 Lightning family — no supervised fine-tuning, reinforcement learning, or distillation has been applied — making it the natural starting point for developers and researchers building their own post-trained models.

The model employs a hybrid Mixture-of-Experts (MoE) architecture, utilizing interleaved Mamba-2 and MoE layers, along with select Attention layers. The model incorporates Multi-Token Prediction (MTP) layers — trained via a dedicated continued pre-training phase — for richer training signals and native speculative decoding, and it is pre-trained using an NVFP4 recipe to maximize compute efficiency. The model has 3B active parameters and 30B parameters in total.

The pre-training corpus spans English, 19 other spoken languages, and 43 programming languages.

This model is ready for commercial and non-commercial use.

What is Nemotron?

NVIDIA Nemotron™ is a family of open models with open weights, training data, and recipes, delivering leading efficiency and accuracy for building specialized AI agents.

License/Terms of Use

Use of this model is governed by the OpenMDW License Agreement, version 1.1 (OpenMDW-1.1).

Benchmarks

Accuracy numbers measured by NVIDIA under a consistent harness (NeMo Gym / Nemo Evaluator SDK); they may differ from vendors' self-reported numbers.

All evaluation results were collected via Nemo Evaluator SDK and NVIDIA's open source container of LM Evaluation Harness except for RULER which uses the NeMo Skills harness. The open source container on LM Evaluation Harness packaged via NVIDIA's Nemo Evaluator SDK used for evaluations can be found here. OSS Evaluation Recipes for Nemotron-3.5 Lightning are available in NeMo Gym.

Deployment Geography: Global

Use Case

This model is intended for developers and researchers building LLMs. As the base checkpoint of the Nemotron 3.5 Lightning family, it is the recommended starting point for pre-training research, continued pre-training on domain corpora, and building custom post-trained variants (SFT, RL, and distillation) via NeMo RL, NeMo Gym, and Megatron-LM.

Release Date

Hugging Face - 08/11/2026

Reference(s)

• NVIDIA Nemotron model family on Hugging Face

Model Architecture

• Architecture Type: Mamba2-Transformer Hybrid Mixture of Experts (MoE) with Multi-Token Prediction (MTP)
• Network Architecture: Nemotron Hybrid MoE
• Number of model parameters: 30B Total / 3B Active

Model Design

The model was pre-trained with over 20T tokens and supports up to 1M context length. The pre-training phase used an NVFP4 recipe. The model includes Multi-Token Prediction (MTP) layers, which predict multiple future tokens to provide richer training signals and enable faster inference via speculative decoding.

Training Methodology

Stage 1: Pre-Training

• NVIDIA-Nemotron-3.5-Lightning-30B-A3B-Base-BF16 model was pre-trained using an NVFP4 recipe with crawled and synthetic code, math, science, and general knowledge data.
• Software used for pre-training: Megatron-LM

Stage 2: Continued Pre-Training for Multi-Token Prediction (MTP)

• The model underwent a continued pre-training phase to train its Multi-Token Prediction (MTP) layers. In this stage, MTP heads learn to predict multiple future tokens, providing richer training signals to the base model and enabling native speculative decoding at inference time.

NVIDIA-Nemotron-3.5-Lightning-30B-A3B-Base-BF16 model is a result of the above work.

Input

• Input Type(s): Text
• Input Format(s): String
• Input Parameters: One-Dimensional (1D): Sequences
• Other Properties Related to Input: Maximum context length up to 1M tokens. The pre-training corpus spans English, 19 other spoken languages, and 43 programming languages.

Output

• Output Type(s): Text
• Output Format: String
• Output Parameters: One-Dimensional (1D): Sequences
• Other Properties Related to Output: Maximum context length up to 1M tokens

Our AI models are designed and optimized to run on NVIDIA GPU-accelerated systems. By leveraging NVIDIA's hardware (e.g. GPU cores) and software frameworks (e.g., CUDA libraries), the model achieves faster training and inference times compared to CPU-only solutions.

Software Integration

• Runtime Engine(s): PyTorch; Megatron-LM; NeMo
• Supported Hardware Microarchitecture Compatibility: NVIDIA Ampere - A100; NVIDIA Blackwell; NVIDIA Hopper - H100-80GB
• Operating System(s): Linux

The integration of foundation and fine-tuned models into AI systems requires additional testing using use-case-specific data to ensure safe and effective deployment. Following the V-model methodology, iterative testing and validation at both unit and system levels are essential to mitigate risks, meet technical and functional requirements, and ensure compliance with safety and ethical standards before deployment.

Model Version(s)

• v1.0 - GA (08/11/2026)

Training, Testing, and Evaluation Datasets:

Training

Data Modality: Text Training Data Size: More than 20 Trillion Tokens Dataset partition: Training [100%], testing [0%], validation [0%] Time period for training data collection: 2013 to December 2025 Time period for testing data collection: 2013 to December 2025 Time period for validation data collection: 2013 to December 2025 Data Collection Method by dataset: Hybrid: Automated, Manually-Collected, Synthetic Labeling Method by dataset: Hybrid: Automated, Manually-Labeled, Synthetic

Properties: NVIDIA-Nemotron-3.5-Lightning-30B-A3B-Base-BF16 is pre-trained on a large corpus of high-quality curated and synthetically-generated data. It is trained in the English language, as well as 19 other spoken languages and 43 programming languages. Our sources cover a variety of document types such as: webpages, dialogue, articles, and other written materials. The corpus spans domains including legal, math, science, finance, and more. We also include a small portion of question-answering, and alignment style data to improve model accuracy. The model was pre-trained for more than 20 trillion tokens.

Base Pre-Training Corpus (Nemotron 3 Foundation)

The foundation of the model is trained on the Nemotron 3 corpus, comprising the following datasets from the Nemotron Pretraining Datasets collection:

Public Datasets

Crawled and Scraped from Online Sources by NVIDIA

The English Common Crawl data was downloaded from the Common Crawl Foundation (see their FAQ for details on their crawling) and includes the snapshots CC-MAIN-2013-20 through CC-MAIN-2025-13. The data was subsequently deduplicated and filtered in various ways described in the Nemotron-CC paper. Additionally, we extracted data for fifteen languages from the following three Common Crawl snapshots: CC-MAIN-2024-51, CC-MAIN-2025-08, CC-MAIN-2025-18. The fifteen languages included were Arabic, Chinese, Danish, Dutch, French, German, Italian, Japanese, Korean, Polish, Portuguese, Russian, Spanish, Swedish, and Thai. As we did not have reliable multilingual model-based quality classifiers available, we applied just heuristic filtering instead—similar to what we did for lower quality English data in the Nemotron-CC pipeline, but selectively removing some filters for some languages that did not work well. Deduplication was done in the same way as for Nemotron-CC.

The GitHub Crawl was collected using the GitHub REST API and the Amazon S3 API. Each crawl was operated in accordance with the rate limits set by its respective source, either GitHub or S3. We collect raw source code and subsequently remove any having a license which does not exist in our permissive-license set.

Private Non-publicly Accessible Datasets of Third Parties

Private Non-publicly Accessible Datasets by NVIDIA

NVIDIA-Sourced Synthetic Datasets

Testing Datasets:

Data Collection Method by dataset

• Hybrid: Automated, Human, Synthetic Labeling Method by dataset
• Hybrid: Automated, Human, Synthetic Properties: This corpus comprises a mix of high-quality standard benchmarks for base language models, covering general knowledge, math, code, commonsense understanding, reading comprehension, multilingual understanding, and long context.

Evaluation Datasets:

Data Collection Method by dataset

• Hybrid: Automated, Human, Synthetic Labeling Method by dataset
• Hybrid: Automated, Human, Synthetic Properties: This corpus comprises a mix of high-quality standard benchmarks for base language models, covering general knowledge, math, code, commonsense understanding, reading comprehension, multilingual understanding, and long context.

Inference

• Acceleration Engine: PyTorch
• Test Hardware:
  • NVIDIA Hopper
    • 1-8x H100
    • 1-8x H200
  • NVIDIA Blackwell
    • GB200

Ethical Considerations

NVIDIA believes Trustworthy AI is a shared responsibility and we have established policies and practices to enable development for a wide array of AI applications. Developers should work with their internal model team to ensure this model meets requirements for the relevant industry and use case and addresses unforeseen product misuse.

We advise against circumvention of any provided safety guardrails contained in the Model without a substantially similar guardrail appropriate for your use case. For more details: Safety and Explainability Subcards.

For more detailed information on ethical considerations for this model, please see the Model Card++ Bias, and Privacy Subcards.

Please report model quality, risk, security vulnerabilities or NVIDIA AI Concerns here.