nvidia · Open weights
Nvidia Nemotron 3 Super 120B
Nvidia's Nemotron 3 Super 120B A12B model from the March 2026 Nemotron 3 release. It uses a hybrid Mamba-Transformer MoE architecture and targets agentic and coding workloads with a 262K context window here.
- Reasoning
- Input
- Text
- Output
- Text
- Discovery tier
- Open weights
Pricing
USD per 1M tokens- Input
- $0.05
- Output
- $0.25
Reference prices as published by the model provider. API example
Specifications
- Context window
- 262.1K
- Max output
- 16.4K
- Average output (7 days)
- 431 tokens
- Longer than 43% of 465 models
Side by side
Compare models
Put Nvidia Nemotron 3 Super 120B next to up to three other models. The comparison stays in the link, so you can share it.
Public evaluations
Benchmarks
Independent evaluation results, with how each score ranks among the models in the Noumen catalog.
Intelligence Index
12.8
Better than 29% of 326 scored models
Indexes
Coding Index
37.7Composite of the coding evaluations
Better than 30% of 168 scored models
Agentic work
AutomationBench-AA
3.8%Workflow automation, penalized for guardrail violations
Better than 21% of 173 scored models
AutomationBench-AA tasks completed
0.3%Workflows finished end to end without a guardrail violation
Better than 19% of 173 scored models
GDPval-AAv2.1
490 EloEconomically valuable occupational tasks, rated head to head
Better than 11% of 187 scored models
Document reasoning
GDP.pdf
2.6%Professional reasoning over PDF documents, all-pass rate
Better than 10% of 171 scored models
AA-LCRv1.1
65.7%Reasoning across long documents
Better than 42% of 309 scored models
Reasoning and knowledge
Humanity’s Last Exam
20.8%Expert-level questions across academic fields
Better than 48% of 323 scored models
GPQA Diamond
80.0%Graduate-level science questions
Better than 48% of 291 scored models
IFBench
71.5%Following precise output instructions
Better than 75% of 215 scored models
Coding
Terminal-Benchv4.0
0.0%Practical software engineering in a terminal
Better than 0% of 171 scored models
SciCode
36.2%Scientific computing in Python
Better than 13% of 174 scored models
Legacy evaluations
Terminal-Bench Hard
28.8%Earlier terminal task suite
Better than 53% of 212 scored models
τ²-Bench Telecom
67.8%Tool use in a customer-support simulation
Better than 52% of 214 scored models
AA-LCR (unversioned)
65.7%Long-context reasoning, earlier grading
Better than 42% of 309 scored models
Scores from Artificial Analysis, updated Oct 26, 2026. Artificial Analysis
Live routing data
Performance and reliability
Recent averages across requests routed for this model. Your own speed depends on prompt length, reasoning effort and load.
- Prompt cache hits
- 1.3%
- Request success (24h)
- 99.9%
- Reliable
Cache hits are the share of input tokens recently served from a prompt cache across all requests. They are not a forecast of your own hit rate.
- Healthy (98% or more)
- Degraded (90–98%)
- Down (under 90%)
- No requests
Developers
Use it through the API
Noumen’s API is OpenAI-compatible: any Chat Completions client works with a base URL and a key.
curl https://www.noumen.xyz/api/v1/chat/completions \
-H "Authorization: Bearer $NOUMEN_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "nvidia/nemotron-3-super-120b-a12b",
"messages": [{ "role": "user", "content": "Explain this stack trace." }]
}'import os
from openai import OpenAI
client = OpenAI(base_url="https://www.noumen.xyz/api/v1", api_key=os.environ["NOUMEN_API_KEY"])
reply = client.chat.completions.create(
model="nvidia/nemotron-3-super-120b-a12b",
messages=[{"role": "user", "content": "Explain this stack trace."}],
)
print(reply.choices[0].message.content)nvidia
Related models
Other nvidia models, starting with the same family.