Skip to main content

Nemotron 3 Super 120B A12B

NVIDIA Nemotron 3 Super is a 120B-parameter open hybrid MoE model, activating just 12B parameters for maximum compute efficiency and accuracy in complex multi-agent applications. Built on a hybrid Mamba-Transformer Mixture-of-Experts architecture with multi-token prediction (MTP), it delivers over 50% higher token generation compared to leading open models. The model features a 1M token context window for long-term agent coherence, cross-document reasoning, and multi-step task planning. Latent MoE enables calling 4 experts for the inference cost of only one, improving intelligence and generalization. Multi-environment RL training across 10+ environments delivers leading accuracy on benchmarks including AIME 2025, TerminalBench, and SWE-Bench Verified. Fully open with weights, datasets, and recipes under the NVIDIA Open License, Nemotron 3 Super allows easy customization and secure deployment anywhere — from workstation to cloud.

Model card on Hugging Face
Input / output price
$0.075 / $0.40 per 1M
Context
262K tokens
Released
Mar 11, 2026
Agent setup

Paste into Claude Code, Cursor, or Codex

Help me set up the Respan gateway: fetch and follow https://www.respan.ai/docs/agent-setup.md. Use model "nemotron-3-super-120b-a12b".

Read docs

Providers

5 providers, cheapest first, OpenRouter last, priced per 1M tokens after gateway discounts.

ProviderInputOutput
Concentrateconcentrate/nemotron-3-120b
$0.075/M$0.40/M
DeepInfradeepinfra/nemotron-3-super-120b-a12b
$0.075/M$0.40/M
Bedrockbedrock/nemotron-3-super-120b-a12b
$0.15/M$0.65/M
Nebiusnebius/nemotron-3-super-120b-a12b
$0.30/M$0.90/M
OpenRouteropenrouter/nemotron-3-super-120b-a12b
$0.075/M$0.40/M

Is a provider down? AWS Bedrock status, OpenRouter status

Pricing

Averaged across 5 providers, in USD per 1M tokens after gateway discounts.

Average input
$0.13 per 1M
Average output
$0.55 per 1M
Average cache read
$0.06 per 1M

Cost by volume

Total cost from 0 to 1T tokens on the cheapest and most expensive provider.

Highest$0.30/MNebiusLowest$0.075/MConcentrate, DeepInfra, and OpenRouter

Throughput and latency

30-day averages from requests through the Respan gateway. Under each figure, every model on the gateway by its average, this one in blue.

Output speed

Higher is better

26tokens/sacross providers

Faster than 54% of the other 595 models on the gateway

Response latency

Lower is better

0.4sacross providers

Faster than 84% of the other 756 models on the gateway

Benchmarks

The Respan Index plus 11 published scores from 3 sources, each linked to where it was published.

19.1/ 100

Respan Index#135 of 176

176 ranked models. The shaded band is this model's 95% interval, 12.6 to 34.2. GPT-4.1 and Qwen3.5-Plus fall inside it, so the gap is within the margin of error.
Nemotron 3 Super 120B A12B by category: Coding 4.4, Agents 17.1 (estimated), Reasoning 23.7, Knowledge & language 17.4, Human preference 40.3
  1. 135NVIDIA Nemotron 3 Super 120B-A12B19.1
  2. 134GPT-4.119.2
  3. 136Qwen3.5-Plus19.0

Published scores

BenchmarkScore
LiveCodeBenchbuild.nvidia.com81.2%
SWE-bench Verifiedbuild.nvidia.com60.5%
WeirdMLHåvard Tveit Ihle38%
Terminal-Bench Core 2.0build.nvidia.com31%

Compare Nemotron 3 Super 120B A12B: NVIDIA Nemotron 3 Super 120B-A12B vs NVIDIA Nemotron 3.5 Super VL 120B-A12B NVFP4, NVIDIA Nemotron 3 Super 120B-A12B vs NVIDIA Nemotron 3 Ultra 550B-A55B NVFP4, NVIDIA Nemotron 3 Super 120B-A12B vs NVIDIA Nemotron 3.5 Lightning 30B-A3B NVFP4, NVIDIA Nemotron 3 Super 120B-A12B vs NVIDIA Nemotron 3 Nano 30B-A3B BF16

Apps

Use Nemotron 3 Super 120B A12B in the tools you already run. Each guide points the app at the Respan gateway.

Setup guides for more apps are in the integration docs.

Nemotron 3 Super 120B A12B FAQ

Short answers, built from the data on this page.

Nemotron 3 Super 120B A12B is a model from NVIDIA, released on Mar 11, 2026. NVIDIA Nemotron 3 Super is a 120B-parameter open hybrid MoE model, activating just 12B parameters for maximum compute efficiency and accuracy in complex multi-agent applications. Built on a hybrid Mamba-Transformer Mixture-of-Experts architecture with multi-token prediction (MTP), it delivers over 50% higher token generation compared to leading open models. The model features a 1M token context window for long-term agent coherence, cross-document reasoning, and multi-step task planning. Latent MoE enables calling 4 experts for the inference cost of only one, improving intelligence and generalization. Multi-environment RL training across 10+ environments delivers leading accuracy on benchmarks including AIME 2025, TerminalBench, and SWE-Bench Verified. Fully open with weights, datasets, and recipes under the NVIDIA Open License, Nemotron 3 Super allows easy customization and secure deployment anywhere — from workstation to cloud.

On the Respan gateway, Nemotron 3 Super 120B A12B costs from $0.075 per 1M input tokens and $0.40 per 1M output tokens on Concentrate, DeepInfra, and OpenRouter, after gateway discounts. Averaged across its 5 providers, it costs $0.13 per 1M input tokens and $0.55 per 1M output tokens.

Nemotron 3 Super 120B A12B is available through 5 providers: Concentrate, DeepInfra, Bedrock, Nebius, and OpenRouter.

Nemotron 3 Super 120B A12B accepts up to 262K tokens of context on its largest provider route.

Point any OpenAI-compatible SDK at https://gateway.respan.ai/v1 with your Respan API key and set the model to "nemotron-3-super-120b-a12b". To set it up with a coding agent, copy the agent setup prompt at the top of this page.

Nemotron 3 Super 120B A12B scores 19.1 out of 100 on the Respan Index v1.1, ranking #135 of 176 models. By category: Coding 4.4, Agents 17.1 (estimated), Reasoning 23.7, Knowledge & language 17.4, and Human preference 40.3. The Respan Index combines independent and lab-reported benchmark results into one score, using benchmark data collected by Respan.

Nemotron 3 Super 120B A12B has 11 published benchmark scores, including LiveCodeBench 81.2%, GPQA (with tools) 82.7%, and AIME 2025 (no tools) 90.2%. Each score links to where it was published; the data is collected by Respan.

Certified where it counts.

Built to meet the security and privacy standards that enterprise and healthcare teams require.

ISO 27001

The internationally recognized standard for information security management.

SOC 2

Secure, compliant management of your data across all of our systems.

GDPR

Operated under GDPR, the world's strictest standard for data privacy.

HIPAA

HIPAA compliant, with a BAA available for healthcare teams.

Works with your entire stack

Use Respan with your favorite frameworks and tools.

AnthropicAI21 LabsxAIVercel AI SDK
LangChainOpenRouterAssemblyAICursor
LiteLLMGroqBasetenMem0
CohereCogneeLlamaIndexFireworks
DeepSeekLinkupHaystackTogether AI
PerplexityInference.netPostHogRubyLLM
SlackMastraAzure OpenAIMistral
AWS BedrockMoonshotHyperspellBAML
ModaInstructorGoogle GeminiParasail
ReductoZapierAgnoNebius
OpenAINovita AIReplicateSuperagent

Built for AI agents.
Break less.
Ship more.