Right this moment, we’re excited to announce the day-zero availability of NVIDIA Nemotron 3 Extremely on Amazon SageMaker JumpStart.<\/p>\n
With this launch, now you can deploy the Nemotron 3 Extremely mannequin utilizing a one-click deployment expertise. Nemotron 3 Extremely is an open mannequin constructed for frontier reasoning and orchestration in long-running autonomous brokers, delivering 5x sooner inference and as much as 30% decrease value for agentic workloads. Nemotron 3 Extremely is optimized for the NVFP4 format, which makes the mannequin a lot sooner and price efficient to host.<\/p>\n
Overview of NVIDIA Nemotron 3 Extremely<\/h2>\n
NVIDIA Nemotron 3 Extremely is an open giant language mannequin with 550 billion whole parameters and 55 billion lively parameters. It’s constructed on a hybrid Transformer-Mamba Combination-of-Specialists (MoE) structure, designed to ship frontier intelligence at a fraction of the compute value of dense fashions of equal high quality.<\/p>\n
Specification
\n Particulars<\/p>\n
Structure
\n Hybrid Transformer-Mamba MoE<\/p>\n
Parameters
\n 550B whole \/ 55B lively<\/p>\n
Context size
\n As much as 1M tokens<\/p>\n
Enter \/ Output
\n Textual content in, textual content out<\/p>\n
Precision
\n NVFP4<\/p>\n
Inference pace
\n 5x sooner for long-running agent workflows<\/p>\n
Price
\n As much as 30% decrease for complicated agentic duties<\/p>\n
![\"\"](\"https:\/\/d2908q01vomqb2.cloudfront.net\/f1f836cb4ea6efb2a0b1b99f41ad8b103eff4b59\/2026\/06\/04\/image-36.png\")
<\/h2>\nWhy agentic AI wants purpose-built fashions<\/h2>\n
Brokers don\u2019t simply reply as soon as. They plan, name instruments, delegate work to sub-agents, test outcomes, and preserve going throughout tons of of turns. Each step provides tokens and compute, so the metrics that matter are job completion at helpful accuracy, time-to-finish, and cost-per-task.<\/p>\n
Nemotron 3 Extremely addresses this immediately. Its MoE structure prompts solely 55B of its 550B parameters per ahead cross, holding throughput excessive even at million-token context lengths. This implies brokers can maintain planning, instrument calling, and self-correction loops that span tons of of turns whereas serving to preserve coherence and handle value.<\/p>\n
Enterprise use circumstances<\/h2>\n
Nemotron 3 Extremely excels in workloads that require sustained multi-step reasoning:<\/p>\n
Agent orchestrators \u2013 coordinate a number of sub-agents, handle state throughout lengthy tool-calling chains
\n Coding brokers \u2013 generate, take a look at, debug, and iterate on code throughout giant repositories
\n Deep analysis \u2013 synthesize data from a number of sources, preserve coherent reasoning over prolonged context
\n Advanced enterprise workflows \u2013 automate multi-step enterprise processes with resolution branching and error restoration<\/p>\n
Getting began with SageMaker JumpStart<\/h2>\n
You may deploy Nemotron 3 Extremely by way of Amazon SageMaker JumpStart with one-click deployment, eradicating the necessity to handle infrastructure or configure serving frameworks.<\/p>\n
Conditions<\/h3>\n
Earlier than you start, ensure you have:<\/p>\n
An AWS account
\n Appropriately scoped permissions for SageMaker JumpStart
\n Ample service quota for GPU situations (for instance, ml.p5en.48xlarge, ml.p5.48xlarge, or ml.g7e.48xlarge)<\/p>\n
Vital: Deploying this mannequin creates a SageMaker endpoint that incurs costs whereas operating. GPU situations like ml.p5en.48xlarge can value a number of {dollars} per hour. See Amazon SageMaker AI pricing for particulars. Keep in mind to delete your endpoint when completed to keep away from ongoing costs.<\/p>\n
Deploy utilizing SageMaker Studio<\/h3>\n
Open Amazon SageMaker Studio
\n Within the left navigation pane, select SageMaker JumpStart
\n Seek for Nemotron 3 Extremely
\n Choose the mannequin card
\n Select Deploy
\n Choose your occasion kind (supported occasion varieties are ml.p5en.48xlarge, ml.p5.48xlarge, or ml.g7e.48xlarge)
\n Evaluate deployment settings (defaults are adequate for many use circumstances)
\n Select Deploy to create the endpoint
\n Anticipate the endpoint standing to indicate InService earlier than continuing to inference<\/p>\n
![\"\"](\"https:\/\/d2908q01vomqb2.cloudfront.net\/f1f836cb4ea6efb2a0b1b99f41ad8b103eff4b59\/2026\/06\/04\/image-37.png\")
<\/p>\n
Deploy utilizing the SageMaker Python SDK<\/h3>\n\n import sagemaker
\nfrom sagemaker.jumpstart.mannequin import JumpStartModel
\nmannequin = JumpStartModel(
\n model_id=”huggingface-reasoning-nvidia-nemotron-3-ultra-550b-a55b-nvfp4″, # Confirm in SageMaker JumpStart mannequin card
\n position=sagemaker.get_execution_role(), # Your SageMaker execution position ARN
\n)
\npredictor = mannequin.deploy(accept_eula=True)\n <\/div>\nRun inference<\/p>\n
\n payload = {
\n “messages”: [{
\n “role”: “user”,
\n “content”: “Break this task into subtasks, identify which tools are needed, and run them in sequence.”
\n }],
\n “max_tokens”: 20480,
\n “temperature”: 0.6,
\n “top_p”: 0.95,
\n}
\nresponse = predictor.predict(payload)
\nprint(response[“choices”][0][“message”][“content”])\n <\/div>\nClear up<\/h2>\n
To keep away from incurring pointless costs, delete the SageMaker endpoint when you find yourself finished:predictor.delete_endpoint()<\/p>\n
Conclusion<\/h2>\n
NVIDIA Nemotron 3 Extremely brings frontier-class reasoning to Amazon SageMaker JumpStart with 5x sooner inference and as much as 30% decrease value for agentic workloads. Its hybrid Transformer-Mamba MoE structure and million-token context window make it purpose-built for the sustained, multi-step reasoning that manufacturing brokers demand.<\/p>\n
Whether or not you’re constructing agent orchestrators, coding brokers, deep analysis techniques, or complicated enterprise automation, Nemotron 3 Extremely is able to deploy as we speak from SageMaker JumpStart.<\/p>\n
Get began now by looking for Nemotron 3 Extremely in Amazon SageMaker JumpStart.<\/p>\n
Concerning the authors<\/h3>\n
![\"\"](\"https:\/\/d2908q01vomqb2.cloudfront.net\/f1f836cb4ea6efb2a0b1b99f41ad8b103eff4b59\/2026\/06\/03\/21170-1.jpeg\")
Dan Ferguson is a Options Architect at AWS, based mostly in New York, USA. As a machine studying companies skilled, Dan works to assist prospects on their journey to integrating ML workflows effectively, successfully, and sustainably.<\/p>\n
![\"\"](\"https:\/\/d2908q01vomqb2.cloudfront.net\/f1f836cb4ea6efb2a0b1b99f41ad8b103eff4b59\/2026\/06\/03\/21170-2.jpeg\")
Malav Shastri is a Software program Improvement Engineer at AWS, the place he works on the Amazon SageMaker JumpStart and Amazon Bedrock groups. His position focuses on enabling prospects to reap the benefits of state-of-the-art open supply and proprietary basis fashions. Malav holds a Grasp\u2019s diploma in Pc Science.<\/p>\n
![\"\"](\"https:\/\/d2908q01vomqb2.cloudfront.net\/f1f836cb4ea6efb2a0b1b99f41ad8b103eff4b59\/2026\/06\/03\/21170-3.jpeg\")
Vivek Gangasani is a Worldwide Chief for Options Structure, SageMaker Inference. He leads Answer Structure, Technical Go-to-Market (GTM) and Outbound Product technique for SageMaker Inference. He additionally helps enterprises and startups deploy and optimize a GenAI fashions and construct AI workflows with SageMaker and GPUs. At present, he’s centered on growing methods and content material for optimizing inference efficiency and use-cases comparable to Agentic workflows, RAG and so forth. In his free time, Vivek enjoys climbing, watching films, and attempting completely different cuisines.<\/p>\n<\/p><\/div>\n
\nfrom sagemaker.jumpstart.mannequin import JumpStartModel
\nmannequin = JumpStartModel(
\n model_id=”huggingface-reasoning-nvidia-nemotron-3-ultra-550b-a55b-nvfp4″, # Confirm in SageMaker JumpStart mannequin card
\n position=sagemaker.get_execution_role(), # Your SageMaker execution position ARN
\n)
\npredictor = mannequin.deploy(accept_eula=True)\n <\/div>\n
Run inference<\/p>\n
\n “messages”: [{
\n “role”: “user”,
\n “content”: “Break this task into subtasks, identify which tools are needed, and run them in sequence.”
\n }],
\n “max_tokens”: 20480,
\n “temperature”: 0.6,
\n “top_p”: 0.95,
\n}
\nresponse = predictor.predict(payload)
\nprint(response[“choices”][0][“message”][“content”])\n <\/div>\n
Clear up<\/h2>\n
To keep away from incurring pointless costs, delete the SageMaker endpoint when you find yourself finished:predictor.delete_endpoint()<\/p>\n
Conclusion<\/h2>\n
NVIDIA Nemotron 3 Extremely brings frontier-class reasoning to Amazon SageMaker JumpStart with 5x sooner inference and as much as 30% decrease value for agentic workloads. Its hybrid Transformer-Mamba MoE structure and million-token context window make it purpose-built for the sustained, multi-step reasoning that manufacturing brokers demand.<\/p>\n
Whether or not you’re constructing agent orchestrators, coding brokers, deep analysis techniques, or complicated enterprise automation, Nemotron 3 Extremely is able to deploy as we speak from SageMaker JumpStart.<\/p>\n
Get began now by looking for Nemotron 3 Extremely in Amazon SageMaker JumpStart.<\/p>\n
Concerning the authors<\/h3>\n
Dan Ferguson is a Options Architect at AWS, based mostly in New York, USA. As a machine studying companies skilled, Dan works to assist prospects on their journey to integrating ML workflows effectively, successfully, and sustainably.<\/p>\n
Malav Shastri is a Software program Improvement Engineer at AWS, the place he works on the Amazon SageMaker JumpStart and Amazon Bedrock groups. His position focuses on enabling prospects to reap the benefits of state-of-the-art open supply and proprietary basis fashions. Malav holds a Grasp\u2019s diploma in Pc Science.<\/p>\n
Vivek Gangasani is a Worldwide Chief for Options Structure, SageMaker Inference. He leads Answer Structure, Technical Go-to-Market (GTM) and Outbound Product technique for SageMaker Inference. He additionally helps enterprises and startups deploy and optimize a GenAI fashions and construct AI workflows with SageMaker and GPUs. At present, he’s centered on growing methods and content material for optimizing inference efficiency and use-cases comparable to Agentic workflows, RAG and so forth. In his free time, Vivek enjoys climbing, watching films, and attempting completely different cuisines.<\/p>\n<\/p><\/div>\n