Sold by: NVIDIA
Deployed on AWS
NVIDIA Nemotron are open models with open weights, training data, and recipes, delivering leading efficiency and accuracy for building specialized AI agents
Overview
Nemotron-3-Super-120B-A12B-FP8 is a large language model (LLM) trained by NVIDIA, designed to deliver strong agentic, reasoning, and conversational capabilities. It is optimized for collaborative agents and high-volume workloads such as IT ticket automation. Like other models in the family, it responds to user queries and tasks by first generating a reasoning trace and then concluding with a final response. The model's reasoning capabilities can be configured through a flag in the chat template.
The model employs a hybrid Latent Mixture-of-Experts (LatentMoE) architecture, utilizing interleaved Mamba-2 and MoE layers, along with select Attention layers. Distinct from the Nano model, the Super model incorporates Multi-Token Prediction (MTP) layers for faster text generation and improved quality, and it is trained using NVFP4 quantization to maximize compute efficiency. The model has 12B active parameters and 120B parameters in total.
The supported languages include: English, French, German, Italian, Japanese, Spanish, and Chinese
This model is ready for commercial use.
Highlights
- Architecture Type: Mamba2-Transformer Hybrid Latent Mixture of Experts (LatentMoE) with Multi-Token Prediction (MTP)
- Network Architecture: Nemotron Hybrid LatentMoE
- Number of model parameters: 120B Total / 12B Active
Details
Sold by
Delivery method
Latest version
Deployed on AWS
New
Introducing multi-product solutions
You can now purchase comprehensive solutions tailored to use cases and industries.
Features and programs
Financing for AWS Marketplace purchases
AWS Marketplace now accepts line of credit payments through the PNC Vendor Finance program. This program is available to select AWS customers in the US, excluding NV, NC, ND, TN, & VT.
Pricing
Pricing is based on actual usage, with charges varying according to how much you consume. Subscriptions have no end date and may be canceled any time.
Additional AWS infrastructure costs may apply. Use the AWS Pricing Calculator to estimate your infrastructure costs.
Dimension | Description | Cost/host/hour |
|---|---|---|
ml.g5.48xlarge Inference (Batch) Recommended | Model inference on the ml.g5.48xlarge instance type, batch mode | $8.00 |
ml.p5.48xlarge Inference (Real-Time) Recommended | Model inference on the ml.p5.48xlarge instance type, real-time mode | $8.00 |
ml.p4de.24xlarge Inference (Real-Time) | Model inference on the ml.p4de.24xlarge instance type, real-time mode | $8.00 |
ml.p5e.48xlarge Inference (Real-Time) | Model inference on the ml.p5e.48xlarge instance type, real-time mode | $8.00 |
ml.p5en.48xlarge Inference (Real-Time) | Model inference on the ml.p5en.48xlarge instance type, real-time mode | $8.00 |
Dimensions summary
You pay by the host hour based on the AWS instance type you run the model on. Pricing is usage-based, so you are charged for the time each instance runs. One option uses the ml.g5.48xlarge instance in batch mode, which processes inference in groups rather than live. The other four options run real-time inference on the ml.p4de.24xlarge, ml.p5.48xlarge, ml.p5e.48xlarge, and ml.p5en.48xlarge instances, which return results on demand. Choose an instance type to match your workload and hardware needs.
Top-of-mind questions for buyers
Both meter by host hour on the instance you run. Batch mode processes inference requests in groups, so you run the ml.g5.48xlarge instance to handle grouped jobs. Real-time mode returns results on demand and uses the ml.p4de, ml.p5, ml.p5e, or ml.p5en instances. You pay for each hour the instance runs.
Charges apply per host hour while the instance runs, whether or not it is actively processing inference. The meter tracks running time, not request volume. To stop software charges, you must shut down the endpoint. Underlying AWS infrastructure fees may apply separately based on instance state.
One host hour is one hour of running a single instance of the chosen type. It covers running the Nemotron 3 Super 120B model on that hardware. Each instance type differs in GPU count and memory, so choose one that matches your workload and throughput needs.
docs.nvidia.com
Helpful?
Vendor refund policy
No Refunds.
Legal
Vendor terms and conditions
Upon subscribing to this product, you must acknowledge and agree to the terms and conditions outlined in the vendor's End User License Agreement (EULA) .
Content disclaimer
Vendors are responsible for their product descriptions and other product content. AWS does not warrant that vendors' product descriptions or other product content are accurate, complete, reliable, current, or error-free.
Delivery details
Amazon SageMaker model
An Amazon SageMaker model package is a pre-trained machine learning model ready to use without additional training. Use the model package to create a model on Amazon SageMaker for real-time inference or batch processing. Amazon SageMaker is a fully managed platform for building, training, and deploying machine learning models at scale.
Deploy the model on Amazon SageMaker AI using the following options:
Version release notes
Additional details
Inputs
- Summary
Input Type(s): Text
Input Format(s): String
Input Parameters: One-Dimensional (1D): Sequences
NVIDIA Nemotron-3-Super-120B-A12B accepts JSON requests via the /invocations API. The request contains a list of messages and optional generation controls. Reasoning behavior is controlled via the chat_template_kwargs parameter, supporting reasoning on, reasoning off, and low-effort reasoning modes. Both streaming (invoke_endpoint_with_response_stream) and non-streaming (invoke_endpoint) are supported.
- Limitations for input type
- Other Properties Related to Input: Maximum context length up to 1M tokens. Supported languages include: English, French, German, Italian, Japanese, Spanish, and Chinese.
- Input MIME type
- application/json
{
"model": "nvidia/nemotron-3-super-120b-a12b",
"messages": [
{
"role": "system",
"content": "You are a helpful assistant."
},
{
"role": "user",
"content": "What is NVIDIA? Answer in 2-3 sentences."
}
],
"temperature": 0.2,
"max_tokens": 512,
"stream": false,
"extra_body": {
"chat_template_kwargs": {
"enable_thinking": true
}
}
}
{
"model": "nvidia/nemotron-3-super-120b-a12b",
"messages": [
{
"role": "system",
"content": "You are a helpful assistant."
},
{
"role": "user",
"content": "What is NVIDIA? Answer in 2-3 sentences."
}
],
"temperature": 0.2,
"max_tokens": 512,
"stream": false,
"extra_body": {
"chat_template_kwargs": {
"enable_thinking": true
}
}
}
Input data descriptions
The following table describes supported input data fields for real-time inference and batch transform.
Field name | Description | Constraints | Required |
|---|---|---|---|
model | Model identifier, must be nvidia/nemotron-3-super-120b-a12b. | String | Yes |
messages | Array of message objects with role (system/user/assistant) and content (String).
| Array | Yes |
temperature | Optional. `temperature=1.0` is recommended for all tasks. | Float | No |
top_p | Optional. `top_p=0.95` is recommended for all tasks. | Float | No |
max_tokens | Optional. Maximum number of tokens to generate. | Integer | No |
extra_body.chat_template_kwargs.enable_thinking | Optional. Enables internal reasoning trace before the final response.
| Boolean | No |
Resources
Support
Vendor support
Free support via NVIDIA NIM Developer Forum:
AWS infrastructure support
AWS Support is a one-on-one, fast-response support channel that is staffed 24x7x365 with experienced and technical support engineers. The service helps customers of all sizes and technical abilities to successfully utilize the products and features provided by Amazon Web Services.
Similar products
Run sensitive AI workloads on shared GPU infrastructure without data exposure. Stained Glass Output Protection protects tokens as they are generated by the LLM. Pairing with Stained Glass Transform (not included in this marketplace product) gives round-trip LLM protection.
Llama 3.3 Nemotron Super 49B V1.5 is a significantly upgraded version of Llama 3.3 Nemotron Super 49B V1 and is a large language model (LLM) which is a derivative of Meta Llama-3.3-70B-Instruct (AKA the reference model). It is a reasoning model that is post trained for reasoning, human chat preferences, and agentic tasks, such as RAG and tool calling. The model supports a context length of 128K tokens.
Llama 3.3 Nemotron Super 49B V1 is a large language model (LLM) which is a derivative of Meta Llama-3.3-70B-Instruct with reasoning capabilities.
NVIDIA Nemotron-4 15B NIM is a 15-billion-parameter large multilingual language model trained on 8T tokens of multilingual, and coding tasks
Nemotron-3-Nano-30B-A3B is a large language model (LLM) trained from scratch by NVIDIA, and designed as a unified model for both reasoning and non-reasoning tasks. It responds to user queries and tasks by first generating a reasoning trace and then concluding with a final response.
Llama 3.1 Nemotron Ultra 253B V1 is a LLM which is a derivative of Meta Llama-3.1-405B-Instruct. It is a reasoning model that is post trained for reasoning, human chat preferences, and tasks, such as RAG and tool calling.
Customer reviews
No customer reviews yet
Be the first to review this product . We've partnered with PeerSpot to gather customer feedback. You can share your experience by writing or recording a review, or scheduling a call with a PeerSpot analyst.