跳轉到主要內容

Nemotron 3 Super 120B A12B (Reasoning)

NVIDIA開源權重NVIDIA Open Model License Agreement · 商用許可

描述

Nemotron 3 Super is a 120B total / 12B active parameter hybrid Mamba-Attention Mixture-of-Experts model optimized for agentic reasoning, coding, planning, tool calling, and long-context analysis. It introduces LatentMoE (projecting tokens into a compressed latent space for expert routing, enabling 4x more experts at the same inference cost), Multi-Token Prediction for native speculative decoding (up to 3x faster generation), and native NVFP4 pretraining on Blackwell. The hybrid architecture interleaves Mamba-2 layers for linear-time sequence processing with strategically placed Transformer attention layers as global anchors, supporting a 1M-token context window. Pre-trained on 25 trillion tokens and post-trained with multi-environment RL across 21 configurations using NeMo Gym/RL with 1.2 million rollouts. Achieves up to 5x higher throughput than previous Nemotron Super and 2.2x higher throughput than GPT-OSS-120B while maintaining comparable accuracy.

發布日期
2026-03-11
參數規模
120.0B
上下文長度
262K
支援模態
text

能力雷達圖

60
general
50
coding
70
reasoning
51
science估算
50
agents
0
multimodal

缺少專門科學評測時,Science 由 LLM Stats 科學得分或推理能力估算。

排行榜排名

暫無排名資料

基準測試分數 (LLM Stats)

(LLM Stats (zeroeval))

Chat

Tau2 Airline56.3%自報
Multi-Challenge55.2%自報

Communication

Tau2 Telecom64.4%自報
Tau2 Retail62.8%自報

General

Arena-Hard v273.9%自報

Instruction Following

IFBench72.6%自報

Language

WMT24++86.7%自報
MMLU-Pro83.7%自報
MMLU-ProX79.4%自報

Long Context

RULER91.8%自報

Math

HMMT 202594.7%自報
AIME 202590.2%自報

Reasoning

GPQANYU + Cohere + Anthropic (2023)82.7%自報
LiveCodeBench81.2%自報
AA-LCR58.3%自報
SWE-Bench Verified53.7%自報
SWE-bench Multilingual45.8%自報
SciCode42.0%自報
Bird-SQL (dev)41.8%自報
BrowseCompOpenAI (2025)31.3%自報
Terminal-Bench 2.0Stanford × Laude Institute (2026)31.0%自報
Terminal-Bench25.8%自報
Humanity's Last Exam22.8%自報

AA 評測指數

(Artificial Analysis)

暫無 AA 評測資料

LLM Stats 分類評分

(LLM Stats (zeroeval))
Language
80
Legal
80
Long Context
80
Finance
80
Healthcare
80
Instruction Following
70
Math
70
Creativity
70
Writing
70
Chat
60
Physics
60
Reasoning
60
General
60
Biology
60
Chemistry
60
Communication
60
Frontend Development
50
Code
50
Tool Calling
50
Search
30
Agents
30
Vision
20

定價

輸入價格$0.3 / 1M tokens
輸出價格$0.9 / 1M tokens
混合價格(3:1)$0.16375 / 1M tokens

速度

Tokens/秒193.7
首Token延遲1.06s
首回答延遲11.38s

供應商價格排行

供應商價格排行

9 個供應商

最便宜: NanoGPT最貴: Nebius Token Factory
供應商輸入輸出
1NanoGPT最便宜
$0.05
$0.25
2Kilo Gateway
$0.06
$0.32
3OpenRouter
$0.075
$0.4
4DeepInfra
$0.085
$0.4
5Requesty
$0.1
$0.5
6Vercel AI Gateway
$0.15
$0.65
7Perplexity Agent
$0.25
$2.5
8NVIDIA主要
$0.3
$0.9
9Nebius Token Factory
$0.3
$0.9

比較該模型在不同 API 供應商之間的定價。

外部連結