Skip to content

NVIDIA · Chat

Nemotron 3 Super 120B A12B

NVIDIA 🇺🇸ChatOpen weights. One of its sources says the weights are publicly available.

Newer version: Nemotron 3.5 Super 120B A12B. Our catalogue has no release date for Nemotron 3.5 Super 120B A12B. We compared the version names in our catalogue.

From Aiml API's listing:
NVIDIA Nemotron 3 Super 120B A12B is the high-capacity MoE variant of the Nemotron 3 family, delivering strong reasoning and long-context performance for demanding text generation and agent workflows while preserving MoE compute efficiency.
llminfo IQ
117.3
Market median price
$0.1725 blended / 1M
Max context
1M
Listings
38

Evidence and sources

🤗 nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-FP8llminfo IQ 117.3, ranked among chat models. From 151 benchmarks. Plenty of data behind it.How llminfo IQ is computedCoding IQ 98.4, 95% interval 79–118. From 11 benchmark families across 5 sources. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.How Coding IQ is computedReleased 10 Mar 2026Maker HQ: US (source)
All rankings: #150 on NanoGPT recently #174 on OpenRouter this week #207 on OpenRouter this month #219 on OpenRouter todaySources: LMArena data, Models.dev data, AI/ML API data, Hugging FacePrimary data from LMArena from Models.dev: name, weights availability from AI/ML API: primary mode, description from Hugging Face: release date
Tags from its sources:open-sourcemoecodingreasoningfunction-calling

Advantages and trade-offs

Compared with 10 chat models, picked mainly by popularity and IQ

Advantages

Open weights available

The weights are public. 3 of the 9 peers that say either way have open weights.

We only make a claim when there's enough data on models of the same type. Updated 22 Nov, 06:07 UTC.Prices come from the model's maker where they sell it, otherwise their official partner, otherwise the market median. This comparison uses 9 maker's prices; 1 market median price.

Input & output

as stated by the model's providers
Text
Text

Capabilities

Supported operations and provider-stated features

ReasoningFunction callingStructured outputPrompt caching

Listings

 Flowdotnvidia/nemotron-3-super-120b-a12b:free · freeSource: FlowDot data262K$0.00$0.00——
 NanoGPTInference by OllamaRegions: UNKnvidia/nemotron-3-super-120b-a12bSource: NanoGPTNot published by the provider$0.0158$0.63——
 Kilonvidia/nemotron-3-super-120b-a12bSources: Kilo AI Gateway data, Models.dev data262K$0.06$0.32——
 OpenRouterInference by DeepInfranvidia/nemotron-3-super-120b-a12b · bf16/flex · bf16Source: OpenRouter data docs262K$0.06$0.32271511ms
 Eden AIInference by DeepInfradeepinfra/nemotron-3-super-120b-a12bSource: Models.dev data262K$0.075$0.40——
 Hugging FaceInference by DeepInfranvidia/NVIDIA-Nemotron-3-Super-120B-A12B-FP8 · fp8Source: Hugging Face Inference Providers data262K$0.075$0.40——
 Nous Portalnvidia/nemotron-3-super-120b-a12bSource: Nous Portal Inference Gateway data262K$0.075$0.40——
 DeepInfranvidia/NVIDIA-Nemotron-3-Super-120B-A12B · bf16Sources: DeepInfra data, LiteLLM data262K$0.075$0.40——
 Kenarinemotron-3-super-120b-a12bSource: Models.dev data262K$0.0971cache read $0.0097$0.48——
 RequestyInference by DeepInfradeepinfra/nvidia/NVIDIA-Nemotron-3-Super-120B-A12BSource: Requesty data262K$0.10$0.50——
Show all 33 listings

Showing 10 of 33: the cheapest per 1M input tokens from each provider.

Also sold by Aihubmix, Crusoe, Doubleword, Eurouter, Nebius AI Studio, Perplexity Agent, Stepping Stone, Thinking Machines, Vercel.

5 listings without a price are hidden.

These providers advertise this model. Researchers have found some resellers serving a different, cheaper model, so see how to check whether a provider meets your needs.

Benchmarks

Named effort levels run from lowest to highest. Select a result to see its source details.

Human preference

Arena ELO38 related benchmarks · 39 results
Arena ELO
Source: LMArena data · Nov 2026
Arena-Hard v2
Reported by NVIDIA · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
NeMo Evaluator / mostly NeMo Skills · Hard Prompt
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
Chinese
Source: LMArena data · Nov 2026
Creative Writing
Source: LMArena data · Nov 2026
English
Source: LMArena data · Nov 2026
Exclude Ties
Source: LMArena data · Nov 2026
Show 32 more resultsHide 32 results
Expert
Source: LMArena data · Nov 2026
French
Source: LMArena data · Nov 2026
German
Source: LMArena data · Nov 2026
Hard Prompts
Source: LMArena data · Nov 2026
Hard Prompts English
Source: LMArena data · Nov 2026
Industry Mathematical
Source: LMArena data · Nov 2026
Longer Query
Source: LMArena data · Nov 2026
Multi Turn
Source: LMArena data · Nov 2026
Non English
Source: LMArena data · Nov 2026

Long context

Reported by NVIDIA · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
NeMo Evaluator / mostly NeMo Skills · published setting
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
RULER5 related benchmarks · 9 results
RULER
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
1M
Reported by NVIDIA Developer · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
official blog Figure 3 · rounded
Reported by NVIDIA · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
dedicated RULER container
128K
Reported by NVIDIA · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
NeMo Evaluator / mostly NeMo Skills · report Table 8 BF16 · dedicated RULER container
256K
Reported by NVIDIA · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
dedicated RULER container
512K
Reported by NVIDIA · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
dedicated RULER container

Coding

LiveCodeBench2 results
Reported by NVIDIA · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
NeMo Evaluator / mostly NeMo Skills · v5 · 2024-07↔2024-12
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
LiveCodeBench v6NeMo Evaluator / mostly NeMo Skills · v6 · 2024-08↔2025-05 · report Table 8 BF16 · Reported by NVIDIA · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
SciCode2 results
Reported by NVIDIA · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
NeMo Evaluator / mostly NeMo Skills · subtask
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
Reported by NVIDIA · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
NeMo Evaluator / mostly NeMo Skills
NeMo Evaluator / mostly NeMo Skills · OpenCode · report Table 8 BF16
Reported by NVIDIA Developer · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
official blog Figure 3 · rounded · harness not labeled
Source: Hugging Face Official Benchmarks data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated

Multilingual

EKI EstonianSource: Keelemudelite mõõdupuu data docs
WMT24++2 results
Reported by NVIDIA · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
NeMo Evaluator / mostly NeMo Skills · en→xx
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated

Agentic

BrowseComp2 results
Source: Steel.dev Agent Leaderboards data · Mar 2026
Setting not stated
Reported by NVIDIA · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
NeMo Evaluator / mostly NeMo Skills · with Search · internal SerpAPI implementation
Tau2Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Terminal-Bench4 related benchmarks · 5 results
Terminal-Bench
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Core 2.0
Reported by NVIDIA · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
NeMo Evaluator / mostly NeMo Skills · Harbor · internal scaffolding
Hard 48
Reported by NVIDIA · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
NeMo Evaluator / mostly NeMo Skills · dedicated open-source container · 48 tasks
Reported by NVIDIA Developer · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
official blog Figure 3 · rounded
Terminal-Bench 2.0
Source: Hugging Face Official Benchmarks data · Reported by the model's developer. The source relaying it did not verify it. claim

Instruction following

IFBench4 results
Reported by NVIDIA · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
NeMo Evaluator / mostly NeMo Skills · prompt · report Table 8 BF16
NeMo Evaluator / mostly NeMo Skills · prompt
Reported by NVIDIA Developer · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
official blog Figure 3 · rounded
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
Scale AI Multi-ChallengeNeMo Evaluator / mostly NeMo Skills · Scale AI multi-turn instruction following · Reported by NVIDIA · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.

Math

AIME 20252 results
Reported by NVIDIA · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
NeMo Evaluator / mostly NeMo Skills · no tools
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
HMMT 20253 related benchmarks · 5 results
HMMT 2025
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
HMMT Feb 25
Reported by NVIDIA · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
NeMo Evaluator / mostly NeMo Skills
Reported by NVIDIA Developer · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
official blog Figure 3 · rounded · tool setting not labeled
Hmmt Feb 2026
Source: Hugging Face Official Benchmarks data · Reported by the model's developer. The source relaying it did not verify it. claim

Knowledge

MMLU-Pro2 results
Reported by NVIDIA · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
NeMo Evaluator / mostly NeMo Skills · published setting
Source: Hugging Face Official Benchmarks data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
MMLU-ProX3 results
Reported by NVIDIA · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
NeMo Evaluator / mostly NeMo Skills · average over languages · report Table 8 BF16
NeMo Evaluator / mostly NeMo Skills · average over languages
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated

Reasoning

Reported by NVIDIA · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
NeMo Evaluator / mostly NeMo Skills · no tools
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
Reported by NVIDIA · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
NeMo Evaluator / mostly NeMo Skills · with tools
Reported by NVIDIA Developer · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
official blog Figure 3 · rounded · +tools

Science

GPQA3 results
Reported by NVIDIA · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
NeMo Evaluator / mostly NeMo Skills
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated

Other

BIRD BenchNeMo Evaluator / mostly NeMo Skills · published setting · Reported by NVIDIA · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Horangi36 related benchmarks · 36 results
4: ALT_환각방지
Source: Horangi Korean Leaderboard 4 data docs
NVIDIA-Nemotron-3-Super-120B-A12B-FP8
4 subset: haerae_bench_v1_rc
Source: Horangi Korean Leaderboard 4 data docs
NVIDIA-Nemotron-3-Super-120B-A12B-FP8
4 subset: haerae_bench_v1_wo_rc
Source: Horangi Korean Leaderboard 4 data docs
NVIDIA-Nemotron-3-Super-120B-A12B-FP8
4 subset: hrm8k
Source: Horangi Korean Leaderboard 4 data docs
NVIDIA-Nemotron-3-Super-120B-A12B-FP8
4 subset: kmmlu
Source: Horangi Korean Leaderboard 4 data docs
NVIDIA-Nemotron-3-Super-120B-A12B-FP8
4 subset: kmmlu_pro
Source: Horangi Korean Leaderboard 4 data docs
NVIDIA-Nemotron-3-Super-120B-A12B-FP8
Show 30 more resultsHide 30 results
4 subset: ko_aime2025
Source: Horangi Korean Leaderboard 4 data docs
NVIDIA-Nemotron-3-Super-120B-A12B-FP8
4 subset: ko_arc_agi
Source: Horangi Korean Leaderboard 4 data docs
NVIDIA-Nemotron-3-Super-120B-A12B-FP8
4 subset: ko_balt_700_semantic
Source: Horangi Korean Leaderboard 4 data docs
NVIDIA-Nemotron-3-Super-120B-A12B-FP8
4 subset: ko_balt_700_syntax
Source: Horangi Korean Leaderboard 4 data docs
NVIDIA-Nemotron-3-Super-120B-A12B-FP8
4 subset: ko_hallulens_nonexistent
Source: Horangi Korean Leaderboard 4 data docs
NVIDIA-Nemotron-3-Super-120B-A12B-FP8
4 subset: ko_hallulens_wikiqa
Source: Horangi Korean Leaderboard 4 data docs
NVIDIA-Nemotron-3-Super-120B-A12B-FP8
4 subset: ko_hle
Source: Horangi Korean Leaderboard 4 data docs
NVIDIA-Nemotron-3-Super-120B-A12B-FP8
4 subset: ko_moral
Source: Horangi Korean Leaderboard 4 data docs
NVIDIA-Nemotron-3-Super-120B-A12B-FP8
4 subset: ko_truthful_qa
Source: Horangi Korean Leaderboard 4 data docs
NVIDIA-Nemotron-3-Super-120B-A12B-FP8
4 subset: kobbq
Source: Horangi Korean Leaderboard 4 data docs
NVIDIA-Nemotron-3-Super-120B-A12B-FP8
4 subset: korean_hate_speech
Source: Horangi Korean Leaderboard 4 data docs
NVIDIA-Nemotron-3-Super-120B-A12B-FP8
4 subset: squad_kor_v1
Source: Horangi Korean Leaderboard 4 data docs
NVIDIA-Nemotron-3-Super-120B-A12B-FP8
4: ALT_유해성방지
Source: Horangi Korean Leaderboard 4 data docs
NVIDIA-Nemotron-3-Super-120B-A12B-FP8
4: ALT_윤리/도덕
Source: Horangi Korean Leaderboard 4 data docs
NVIDIA-Nemotron-3-Super-120B-A12B-FP8
4: ALT_제어성
Source: Horangi Korean Leaderboard 4 data docs
NVIDIA-Nemotron-3-Super-120B-A12B-FP8
4: ALT_편향성방지
Source: Horangi Korean Leaderboard 4 data docs
NVIDIA-Nemotron-3-Super-120B-A12B-FP8
4: FINAL_SCORE
Source: Horangi Korean Leaderboard 4 data docs
NVIDIA-Nemotron-3-Super-120B-A12B-FP8
4: GLP_구문해석
Source: Horangi Korean Leaderboard 4 data docs
NVIDIA-Nemotron-3-Super-120B-A12B-FP8
4: GLP_논리적추론
Source: Horangi Korean Leaderboard 4 data docs
NVIDIA-Nemotron-3-Super-120B-A12B-FP8
4: GLP_수학적추론
Source: Horangi Korean Leaderboard 4 data docs
NVIDIA-Nemotron-3-Super-120B-A12B-FP8
4: GLP_의미해석
Source: Horangi Korean Leaderboard 4 data docs
NVIDIA-Nemotron-3-Super-120B-A12B-FP8
4: GLP_일반적지식
Source: Horangi Korean Leaderboard 4 data docs
NVIDIA-Nemotron-3-Super-120B-A12B-FP8
4: GLP_전문적지식
Source: Horangi Korean Leaderboard 4 data docs
NVIDIA-Nemotron-3-Super-120B-A12B-FP8
4: GLP_정보검색
Source: Horangi Korean Leaderboard 4 data docs
NVIDIA-Nemotron-3-Super-120B-A12B-FP8
4: GLP_추상적추론
Source: Horangi Korean Leaderboard 4 data docs
NVIDIA-Nemotron-3-Super-120B-A12B-FP8
4: GLP_코딩능력
Source: Horangi Korean Leaderboard 4 data docs
NVIDIA-Nemotron-3-Super-120B-A12B-FP8
4: GLP_표현
Source: Horangi Korean Leaderboard 4 data docs
NVIDIA-Nemotron-3-Super-120B-A12B-FP8
4: GLP_함수호출
Source: Horangi Korean Leaderboard 4 data docs
NVIDIA-Nemotron-3-Super-120B-A12B-FP8
4: 가치정렬성능(ALT)_AVG
Source: Horangi Korean Leaderboard 4 data docs
NVIDIA-Nemotron-3-Super-120B-A12B-FP8
4: 범용언어성능(GLP)_AVG
Source: Horangi Korean Leaderboard 4 data docs
NVIDIA-Nemotron-3-Super-120B-A12B-FP8
PinchBench3 related benchmarks · 3 results
v2 · Best Success Rate
Reported by PinchBench · via BenchBoard docs
OpenClaw · PinchBench v2.0.0 · task count: 147 on live leaderboard / 148 in v2.0.0 release notes · best run
v2 · Average Success Rate
Reported by PinchBench · via BenchBoard docs
OpenClaw · PinchBench v2.0.0 · task count: 147 on live leaderboard / 148 in v2.0.0 release notes · average across 15 runs
Version/Metric Unspecified
Reported by NVIDIA Developer · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
official blog prose · OpenClaw · version and metric not specified
TAU2Bench2 results
Reported by NVIDIA · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
NeMo Evaluator / mostly NeMo Skills · Airline/Retail/Telecom average · default prompt
Reported by NVIDIA Developer · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
official blog Figure 3 · rounded · aggregate
τ²-benchReported by NVIDIA · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.

Data credits: BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links. · Data by LLM Stats · Hugging Face official benchmark leaderboards; licence varies by benchmark · SEA-HELM leaderboard, AI Singapore (AI Products team), National University of Singapore. Published aggregate scores; results licence not stated (SEA-HELM code is MIT). · LMArena, CC BY 4.0 (licence) · Baromètre de l’écriture IA, Johannes Eva; public numeric ranking data, © 2026 Johannes Eva, no results licence stated · Steel.dev Agent Leaderboards, MIT (licence) · BullshitBench by Peter Gostev, MIT (licence) · Keelemudelite mõõdupuu: Independent LLM leaderboard for Estonian. Eesti Keele Instituut. https://moodupuu.eki.ee, CC BY 4.0. Latest per-benchmark results selected with partial-run flags retained. (licence) · OpenEvals/leaderboard-data, MIT (licence) · Horangi 4, Weights & Biases Korea; public publisher-selected numeric evaluation results · Swallow LLM Leaderboard v2, Swallow LLM team (Institute of Science Tokyo). Contents CC BY 4.0; Japanese post-trained task scores on the source's 0..1 scale. (licence)

Also known as: deepinfra/nemotron-3-super-120b-a12b, deepinfra/nvidia/nvidia-nemotron-3-super-120b-a12b, nebius/nvidia/nemotron-3-super-120b-a12b, nemotron-3-super-120b-a12b-free, nvidia-nemotron-3-super-120b-a12b, nvidia/nemotron-3-super-120b-a12b, nvidia/nemotron-3-super-120b-a12b:free, nvidia/nemotron-3-super-120b-a12b:thinking, nvidia/nvidia-nemotron-3-super-120b-a12b, nvidia/nvidia-nemotron-3-super-120b-a12b-bf16, nvidia/nvidia-nemotron-3-super-120b-a12b-fp8, nvidia/nvidia-nemotron-3-super-120b-a12b-nvfp4, nvidia_nim/nvidia/nemotron-3-super-120b-a12b

Like this one. Tap or hover any dotted value, label or icon to see where it comes from and how it is measured. open details when tapped or hovered. How this data works