The pre-training data has a cutoff date of September 2025.
The post-training data has a cutoff date of May 2026.
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.
Description
NVIDIA-Nemotron-3.5-Lightning-30B-A3B-NVFP4 is a large language model (LLM) trained by NVIDIA.
The model employs a hybrid Mixture-of-Experts architecture, utilizing interleaved Mamba-2 and MoE layers, along with select Attention layers. The Lightning 3.5 model is released alongside a number of speculative decoding methods for faster text generation. The model has 3B active parameters and 30B parameters in total.
This model is ready for commercial use.
Quick Start
To get quickly started on DGX Spark (GB10) you can use the following command.
We evaluated our model on the following benchmarks:
Task
Nemotron-3.5-Lightning-30B-A3B-BF16
Nemotron-3.5-Lightning-30B-A3B-NVFP4
General Knowledge
MMLU Pro
81.94
81.62
AA-Omniscience
17.50
16.63
Reasoning
GPQA Diamond (no tools)
75.44
75.57
HLE (text-only, no tools)
11.72
10.47
SciCode
32.60
31.38
Coding & Agentic
SWE-bench Verified
51.56
52.80
SWE-bench Multilingual
39.33
36.47
Terminal-Bench 2.1
24.58
23.46
PinchBench
85.37
83.43
BrowseComp
36.97
36.81
τ³-bench (Banking)
9.28
9.48
GDPval-AA-V2
832
865
Instruction Following
IFBench (loose)
71.88
72.88
Long Context
AA-LCR
52.00
49.19
Accuracy numbers measured by NVIDIA under a consistent harness (NeMo Gym / Nemo Evaluator SDK); they may differ from vendors' self-reported numbers.
For reproducibility, the evaluation recipes, installation instructions, and commands for NVIDIA Nemotron 3.5 Lightning were collected and published in NeMo Gym. The reported results cover the release evaluation suite, including knowledge and reasoning, instruction following, coding, agentic, tool-use, and long-context. Most evaluations use NeMo Gym-native harnesses while a small subset, including SWE-Bench and Terminal-Bench, used NeMo Evaluator natively. The published recipes specify the benchmark-specific containers, prompts, inference parameters, parser configurations, and scoring settings used to produce the results.
These numbers were measured with and apply to the official NVFP4 checkpoint
Agentic Coding Benchmarks
Additional harness-level coding-agent results for SWE-Bench Verified and Terminal-Bench 2.1 are shown below.
Agentic Coding Benchmarks
Deployment Geography: Global
Use Case
NVIDIA-Nemotron-3.5-Lightning-30B-A3B-NVFP4 is a general purpose reasoning and chat model intended to be used in English and coding languages. Other non-English languages (Spanish, French, German, Italian, Japanese) are also supported. Intended for developers designing AI Agent systems, chatbots, RAG systems, and other AI-powered applications. Also suitable for typical instruction-following tasks.
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.
Training Methodology
Stage 1: Pre-Training
NVIDIA-Nemotron-3.5-Lightning-30B-A3B-NVFP4 model was pre-trained using an NVFP4 recipe with crawled and synthetic code, math, science, and general knowledge data.
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. This phase aligns the MTP layers with the base model's distribution.
Stage 3: Supervised Fine-Tuning
The model was further fine-tuned on synthetic code, math, science, tool calling, instruction following, structured outputs, and general knowledge data. This stage incorporated data designed to support long-range retrieval and multi-document aggregation.
Stage 4: Reinforcement Learning
The model underwent multi-environment reinforcement learning using GRPO (Group Relative Policy Optimization) across math, code, science, instruction following, multi-step tool use, multi-turn conversations, and structured output environments. It utilized an asynchronous RL architecture that decouples training from inference and leverages MTP to accelerate rollout generation.
We performed post-training quantization (PTQ) with Nvidia Model Optimizer using the following recipe: Four Over Six NVFP4 (a variant of static MSE calibration) W4A16 on routed and shared experts, FP8 per-tensor dynamic scales on mamba in_proj/out_proj and KV cache. We used a subset of the Nemotron Ultra validation set for calibration with 1000 samples at 32k token length.
NVIDIA-Nemotron-3.5-Lightning-30B-A3B-NVFP4 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. Supported languages include English, Spanish, French, German, Italian, and Japanese.
Other Properties Related to Output: Maximum context length up to 1M tokens
Our AI models are designed and/or 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.
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.
Lightning 3.5 ships with two external draft models for speculative decoding as well as MTP (Multi-Token Prediction). While DSpark is our recommended default (see Choose your deployment), some workloads are better served without speculative decoding, or by DFlash or MTP:
DSpark: A semi-autoregressive speculative-decoding drafter that proposes a whole block of candidate tokens in a single forward pass from a parallel backbone. This is recommended for DGX Spark, as well as low-concurrency data centre deployments.
DFlash: A speculative-decoding drafter that uses a lightweight block-diffusion model to generate an entire draft block in one forward pass.
MTP: A modeling technique that trains the network to predict several future tokens at each position instead of only the next one.
DSpark is the recommended default for DGX Spark and latency-sensitive, low-concurrency serving. For maximum-throughput batch serving on H100, no speculative decoding is fastest.
vLLM
For more indepth instructions on how to deploy through vLLM, head here
For max throughput deployments, use the following configuration, no speculative decoding strategy is best for this serving configuration, and due to memory constraints the Mamba cache dtype is set as FP16:
Context Length: The H100 and GB200 snippets above serve the model's full 1M-token context window by default. If you're memory-constrained — or want more KV-cache headroom at high concurrency — lower --max-model-len to match your workload.
TensorRT-LLM
For more indepth instructions on how to deploy through TensorRT-LLM, head here
The blocks above run the balanced baseline. To enable one of the speculative decoding strategies described above, append the matching flags to any command:
MTP — the draft head is embedded in the target checkpoint (no separate download):
Context length: The commands above serve the model's full context window by default. Set --context-length to a smaller value if you're memory-constrained or want more KV-cache headroom at higher concurrency.
Local AI (RTX 5090, DGX Spark, and RTX 6000 Pro)
The following recipes are validated and provided by our fantastic partners.
Ollama
ollama version: 0.32.9
For local, single-command use, Nemotron 3.5 Lightning is on Ollama, with tool-calling and thinking enabled.
ollama run nemotron-3.5-lightning
Validated context: Ollama will dynamically set context based on available VRAM (under 24GB -> 4K, 24-48GB -> 32K, 48GB+ -> 256K). You can use /set parameter num_ctx DESIRED_CONTEXT_LENGTH which may result in CPU offloading for higher context limits.
Validated context: the examples set -c 40960 (~40K); raise it as VRAM allows.
LM Studio
You can also run Nemotron 3.5 Lightning 30B A3B through LM Studio from this resource!
API Client
The examples below use the OpenAI-compatible client and work with any of the serving backends above. All backends serve on port 8000 (vLLM and TRT-LLM by default; SGLang via --port 8000), so the base_url works as-is. Recommended sampling settings are Temperature 1.0 and Top_P 0.95.
The vLLM snippets above pass the checkpoint to --model $MODEL_CKPT, and vLLM registers the model under that same identifier (nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-NVFP4) — none of the commands set --served-model-name. For the other backends — or if you add a --served-model-name of your own — copy the identifier returned by GET /v1/models into MODEL below.
Lightning 3.5 exposes reasoning control through chat-template kwargs: thinking enabled (the default), and thinking disabled for direct answers.
Reasoning ON / OFF and streaming examples: Click to expand!
Reasoning ON (default)
python
1response = client.chat.completions.create(2 model=MODEL,3 messages=[{"role":"user","content":"Write a haiku about GPUs"}],4 max_tokens=16000,5 temperature=1.0,6 top_p=0.95,7 extra_body={"chat_template_kwargs":{"enable_thinking":True}}8)9print(response.choices[0].message.content)
Reasoning OFF
python
1response = client.chat.completions.create(2 model=MODEL,3 messages=[{"role":"user","content":"What is the capital of Japan?"}],4 max_tokens=16000,5 temperature=1.0,6 top_p=0.95,7 extra_body={"chat_template_kwargs":{"enable_thinking":False}}8)9print(response.choices[0].message.content)
Streaming
python
1stream = client.chat.completions.create(2 model=MODEL,3 messages=[{"role":"user","content":"Explain speculative decoding in two sentences"}],4 max_tokens=16000,5 temperature=1.0,6 top_p=0.95,7 stream=True,8)9for chunk in stream:10print(chunk.choices[0].delta.content or"", end="", flush=True)
Tool Calling
The TRT-LLM snippet above already launches with the required parsers (--reasoning_parser nemotron-v3 --tool_parser qwen3_coder). For vLLM, add the following to any serve command above:
NOTE: For coding agents, add extra_body={"chat_template_kwargs": {"force_nonempty_content": True}} to the API call, as shown below.
python
1tools =[{2"type":"function",3"function":{4"name":"get_weather",5"description":"Get the current weather for a city",6"parameters":{7"type":"object",8"properties":{"city":{"type":"string"}},9"required":["city"],10},11},12}]1314response = client.chat.completions.create(15 model=MODEL,16 messages=[{"role":"user","content":"What's the weather in Santa Clara?"}],17 tools=tools,18 max_tokens=16000,19 temperature=1.0,20 top_p=0.95,21 extra_body={"chat_template_kwargs":{"force_nonempty_content":True}},22)23print(response.choices[0].message.tool_calls)
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
NVIDIA-Nemotron-3.5-Lightning-30B-A3B-NVFP4 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.
The post-training corpus for NVIDIA-Nemotron-3.5-Lightning-30B-A3B-NVFP4 consists of high-quality curated and synthetically-generated data. Primary languages used for post-training include English, French, German, Italian, Japanese, Spanish, and Chinese.
These datasets, such as FinePDFs, EssentialWeb, HotpotQA, SQuAD, and HelpSteer3, do not collectively or exhaustively represent all demographic groups (and proportionally therein). For instance, these datasets do not contain explicit mentions of demographic classes such as age, gender, or ethnicity in 64-99% of samples, depending on the source. In the subset where such terms are present, document-based datasets (FinePDFs and EssentialWeb) contain representational skews, such as references to "male" outnumbering those to "female", and mentions of "White" as the most frequent among ethnic identifiers (comprising 43-44% of ethnicity mentions). To mitigate these imbalances, we recommend considering evaluation techniques such as bias audits, fine-tuning with demographically balanced datasets, and mitigation strategies like counterfactual data augmentation to align with the desired model behavior. This evaluation used a 3,000-sample subset per dataset, identified as the optimal threshold for maximizing embedder accuracy.
During post-training, we generate synthetic data by distilling trajectories, solutions, and translations from strong teacher models and agent systems, often grounded in real tasks or documents and aggressively filtered for quality. For math, code, and science, we start from curated problem sets and use open source permissive models such as GPT-OSS-120B to produce step-by-step reasoning traces, candidate solutions, best-of-n selection traces, and verified CUDA kernels. For long-context and science, we build synthetic QA and reasoning data by retrieving passages from long documents, generating MCQ/OpenQA questions and answers, and paraphrasing them into multiple prompt/response formats to ensure diversity. Across all pipelines we stack automated verification—compilers, numerical checks, language identification—to ensure our data is high quality.
For all domains, we apply a unified data filtering pipeline to ensure that only high-quality, license-compliant, and verifiable samples are used for post-training. We first discard malformed examples using structural checks (e.g., missing tool definitions when tool calls are present). We then aggressively filter reasoning traces exhibiting pathological repetition, such as repeated n-grams within a sliding window or across the entire trajectory, which we found to be a strong indicator of malformed or low-quality reasoning. Finally, based on internal audits of synthetically generated datasets, we observed that some teacher models occasionally produce reasoning traces and final responses that implicitly align with specific political entities or promote nationalistic narratives. To mitigate this, we apply targeted keyword- and regex-based filters and remove all trajectories matching such behavior.
Alongside the model, we release our final pre-training and post-training data, as outlined in this section. For ease of analysis, there is a sample set that is ungated. For all remaining code, math and multilingual data, gating and approval is required, and the dataset is permissively licensed for model training purposes.
A massive collection of English web data filtered from Common Crawl, including 2.5T+ tokens of new organic, translated, and synthetically rephrased content.
Nemotron-CC-Code-v1
427.9B
High-quality code tokens extracted from Common Crawl using the Lynx + LLM pipeline to preserve structure and equations.
Nemotron-Pretraining-Code-v1 & v2 & v3
1.7T
Curated GitHub code references with multi-stage filtering, deduplication, and large-scale synthetic code data.
Nemotron-CC-Math-v1
133.3B
High-quality math pre-training dataset preserving LaTeX formatting and mathematical structures.
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.
Dataset
Modality
Dataset Size
Collection Period
Collecting Organisation
English Common Crawl
Text
3.36T
4/8/2025
NVIDIA Advanced Deep Learning Research
English Common Crawl 1.1
Text
Not disclosed
10/2/2025
NVIDIA Advanced Deep Learning Research
Multilingual Common Crawl
Text
812.7B
5/1/2025
NVIDIA Advanced Deep Learning Research
GitHub Crawl
Text
747.4B
4/29/2025
NVIDIA Advanced Deep Learning Research
GitHub Crawl 1.1
Text
172.7B
9/30/2025
NVIDIA Advanced Deep Learning Research
Private Non-publicly Accessible Datasets of Third Parties
Dataset
Model(s) used
Global Regulation
Unknown
TAUS Translation Memory
Unknown
Scale HLE
Unknown
HackerRank Coding
Unknown
RL data for Search
Gemini 3; GPT-5
Mercor SWE-AgentsV1
Undisclosed
Private Non-publicly Accessible Datasets by NVIDIA
Dataset
Model(s) used
Simple Minesweeper
Undisclosed
Simple Sudoku
Undisclosed
Multitool Typewriter Hard
Undisclosed
Machine Translation of News Commentary and TAUS Translation Memory
CommonPile (caselaw_access_project_filtered); California Code of Regulations; Judicial Ethics Opinions; GLOBALCIT; CUAD; Nemotron Personas; ToSDR Terms of Service Corpus; CodeHima/TOS_Dataset; ContractNLI; CaseHOLD; Code of Federal Regulations; Canadian Case Law (subsets that allow commercial use)
Synthetic Structured Outputs from Qwen3-30B-A3B-Instruct-2507, Qwen3-30B-A3B-Thinking-2507, Qwen3-235B-A22B-Instruct-2507, and Qwen3-235B-A22B-Thinking-2507
[doubtnut]; [Pile-FreeLaw]; [Llama Nemotron Dataset]; [askfilo]; [EssentialAI/essential-web-v1.0]; [Vedantu]; [auxiliary_train]; [cdquestions.com]; [AMC 8 Problems and Solutions, AMC 10 Problems and Solution, and AIME Problems and Solutions]; [AAPT]; [ICHO-IPH0 Dataset]; [LIMO dataset (Less is More for Reasoning)]
Synthetic Verilog and SystemVerilog Code from DeepSeek-R1-0528 and GPT-OSS-120B
Text
1,233,247
[Verilog/SystemVerilog seed code]
[SDR: DeepSeek R1 0528 and GPT-OSS-120B]; [Filtering: Claude 4 Sonnet]
Synthetic Aider Python Tasks from DeepSeek-R1-0528
Text
236,099
[Exercism (GitHub Python)]
[Deepseek R1 0528]
Synthetic Chat Reasoning-Off Data from GLM-5
Text
646,738
[lmarena-ai/repochat-arena-preference-4k user prompts]
[Multi-turn conversations generated by GLM-5 with best-of-4 selection via Qwen3-Nemotron-235B-A22B-GenRM]
Synthetic Chat Reasoning-On Data from GLM-5
Text
644,286
[lmarena-ai/repochat-arena-preference-4k user prompts]; [lmarena-ai/arena-expert-5k user prompts]; [lmarena-ai/arena-human-preference-55k user prompts]; [lmarena-ai/arena-human-preference-100k user prompts]; [lmarena-ai/arena-human-preference-140k user prompts]
[Multi-turn conversations generated by GLM-5 with best-of-4 selection via Qwen3-Nemotron-235B-A22B-GenRM]
Synthetic Multilingual Safety from Riva-Translate-4B-Instruct-v1.1
Text
132,067
[Safety SFT Data: Ultra]
[nvidia/Riva-Translate-4B-Instruct-v1.1]
Synthetic Science Reasoning Effort Medium
Text
502,722
[science-reasoning-effort-medium-v0]
Undisclosed
Synthetic Telecom Tool-Use Trajectories from gpt-oss-120b
Text
12,455
[Existing Tau2 telecom trajectories originally generated with DeepSeek V3.2]
[gpt-oss-120b]
Synthetic Terminal Bench Data from OpenReasoningv2
Synthetic STEM from Qwen3-235B-A22B-Instruct-2507 and gpt-oss-120b
Text
Undisclosed
[arXiv]; [National Institutes of Health ExPorter]; [BioRxiv]; [PMC Article]; [USPTO Backgrounds]; [peS2o]; Global Regulation; [CORE]; [PG-19]; [DOAB CC BY & CC BY-SA subset]; [NDLTD]
[Qwen3-235B-A22B-Instruct-2507]; [gpt-oss-120b]
Translation Data from TAUS
Text
1,618,055
[TAUS proprietary dataset]
Undisclosed
Synthetic Art of Problem Solving and Stack Exchange from gpt-oss-120b, Qwen2.5-32B-Instruct, and Goedel-Prover-V2-32B
Text
860,469
[Nemotron-Math-Proofs-v1]
[Goedel-Prover-V2-32B]
Synthetic Art of Problem Solving and Stack Exchange from gpt-oss-120b
Text
1,201,815
[Upstream released math dataset]; [AoPS]; [StackOverflow / StackExchange]
[gpt-oss-120b]
Synthetic Multilingual Science and Code data from DeepSeek-R1, DeepSeek-R1-0528, Qwen2.5-32B-Instruct, and Qwen3-235B-A22B, translated with Qwen2.5-32B-Instruct and Qwen2.5-14B-Instruct
Synthetic Multilingual Science and Code data from DeepSeek-R1, DeepSeek-R1-0528, Qwen2.5-32B-Instruct, and Qwen3-235B-A22B, translated with Qwen2.5-32B-Instruct and Qwen2.5-14B-Instruct (Stack Exchange lineage)
Synthetic Science MCQ and QA Diversity from GPT-OSS and Kimi-K2
Text
30,358
[doubtnut]; [Pile-FreeLaw]; [Llama Nemotron Dataset]; [askfilo]; [EssentialAI/essential-web-v1.0]; [Vedantu]; [auxiliary_train]; [cdquestions.com]; [AMC 8 Problems and Solutions, AMC 10 Problems and Solution, and AIME Problems and Solutions]; [AAPT]; [ICHO-IPH0 Dataset]; [LIMO dataset (Less is More for Reasoning)]
[GPT-OSS]; [Kimi-K2]
Synthetic Science HLE with Python from GPT-OSS and Kimi-K2
Text
85,184
[doubtnut]; [Pile-FreeLaw]; [Llama Nemotron Dataset]; [askfilo]; [EssentialAI/essential-web-v1.0]; [Vedantu]; [auxiliary_train]; [cdquestions.com]; [AMC 8 Problems and Solutions, AMC 10 Problems and Solution, and AIME Problems and Solutions]; [AAPT]; [ICHO-IPH0 Dataset]; [LIMO dataset (Less is More for Reasoning)]
[GPT-OSS]; [Kimi-K2]
Synthetic Science Search and Python from GPT-OSS and Kimi-K2
Text
6,179
[doubtnut]; [Pile-FreeLaw]; [Llama Nemotron Dataset]; [askfilo]; [EssentialAI/essential-web-v1.0]; [Vedantu]; [auxiliary_train]; [cdquestions.com]; [AMC 8 Problems and Solutions, AMC 10 Problems and Solution, and AIME Problems and Solutions]; [AAPT]; [ICHO-IPH0 Dataset]; [LIMO dataset (Less is More for Reasoning)]
[GPT-OSS]; [Kimi-K2]
Synthetic Science Search from GPT-OSS and Kimi-K2
Text
32,554
[doubtnut]; [Pile-FreeLaw]; [Llama Nemotron Dataset]; [askfilo]; [EssentialAI/essential-web-v1.0]; [Vedantu]; [auxiliary_train]; [cdquestions.com]; [AMC 8 Problems and Solutions, AMC 10 Problems and Solution, and AIME Problems and Solutions]; [AAPT]; [ICHO-IPH0 Dataset]; [LIMO dataset (Less is More for Reasoning)]
[GPT-OSS]; [Kimi-K2]
Synthetic Finance Reasoning from GPT-OSS-120B and Qwen3-235B-A22B-Instruct-2507
Text
326,700
[SEC filings]
[GPT-OSS-120B, Qwen3-235B-A22B-Instruct-2507]
Synthetic Science Diversity MCQ from GPT-OSS and Kimi-K2
Text
532,942
[doubtnut]; [Pile-FreeLaw]; [Llama Nemotron Dataset]; [askfilo]; [EssentialAI/essential-web-v1.0]; [Vedantu]; [auxiliary_train]; [cdquestions.com]; [AMC 8 Problems and Solutions, AMC 10 Problems and Solution, and AIME Problems and Solutions]; [AAPT]; [ICHO-IPH0 Dataset]; [LIMO dataset (Less is More for Reasoning)]
[GPT-OSS]; [Kimi-K2]
Synthetic Science Diversity OpenQ from GPT-OSS and Kimi-K2
Text
131,045
[doubtnut]; [Pile-FreeLaw]; [Llama Nemotron Dataset]; [askfilo]; [EssentialAI/essential-web-v1.0]; [Vedantu]; [auxiliary_train]; [cdquestions.com]; [AMC 8 Problems and Solutions, AMC 10 Problems and Solution, and AIME Problems and Solutions]; [AAPT]; [ICHO-IPH0 Dataset]; [LIMO dataset (Less is More for Reasoning)]
[GPT-OSS]; [Kimi-K2]
Synthetic Science Reasoning No-Tool from GPT-OSS and Kimi-K2
Text
2,085,600
[doubtnut]; [Pile-FreeLaw]; [Llama Nemotron Dataset]; [askfilo]; [EssentialAI/essential-web-v1.0]; [Vedantu]; [auxiliary_train]; [cdquestions.com]; [AMC 8 Problems and Solutions, AMC 10 Problems and Solution, and AIME Problems and Solutions]; [AAPT]; [ICHO-IPH0 Dataset]; [LIMO dataset (Less is More for Reasoning)]
Synthetic Nemotron-Personas-USA from gpt-oss-120b and Qwen3-8B
Text
Undisclosed
[Nemotron-Personas-USA]
[gpt-oss-120b]; [Qwen3-8B]
Vendor Terminal Bench-like Tasks (Droid)
Text
Undisclosed
[Droid Harness Pivot vendor data]
Undisclosed
Language Distribution in Post-Training
For our post-training recipe, we focused on the following languages in addition to English: French, German, Italian, Japanese, Spanish, and Chinese. Those languages were represented in the form of multilingual reasoning and translation tasks.
Testing Datasets:
Data Collection Method by dataset
Hybrid: Automated, Manually-Collected, Synthetic
Labeling Method by dataset
Hybrid: Automated, Manually-Labeled, Synthetic
Properties: This corpus comprises a mix of high-quality standard benchmarks and test suites for modern agentic AI. These benchmarks test model capabilities on tasks such as tool-calling and instruction following.
Evaluation Datasets:
Data Collection Method by dataset
Hybrid: Automated, Manually-Collected, Synthetic
Labeling Method by dataset
Hybrid: Automated, Manually-Labeled, Synthetic
Properties: This corpus comprises a mix of high-quality standard benchmarks and test suites for modern agentic AI. These benchmarks test model capabilities on tasks such as tool-calling and instruction following.
Inference
Acceleration Engine: PyTorch
Test Hardware:
NVIDIA Hopper
1-8x H100
1-8x H200
NVIDIA Blackwell
GB200
DGX Spark (GB10)
GeForce RTX 5090
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.
