The LLM benchmark aggregator.

Nemotron 3 Super

NVIDIA · 124B · open weights

Strongest in Core abilities (#153 of 204, on 1 of its 3 boards), weakest in General intelligence (#158 of 204). Above par in 0 of 13 scopes. Among the models it meets almost everywhere, it finishes behind GPT-5.6 Terra and K2 Horizon 375B A23B and ahead of Llama 4 Maverick Instruct and DeepSeek V3.

§ 1 · Profile

What it is good at

Bars run from 50 — the average of the models each source lists — so right of the line is above par. The middle column is the gap to whoever leads that scope.

Core abilities44.6−22.8#153/2041/3

General intelligence42.2−27.9#158/2041/3

Long context40−24.6—/0✱✱

Agents46.9−21.2—/267✱0/5

Tool use42.1−31.9—/81✱0/1

Reasoning49−17.9—/178✱0/4

Science49.6−13.9—/122✱0/1

Expert reasoning43.9−17—/0✱✱

Coding49.7−19.4—/164✱0/5

Agentic coding49.4−19.3—/157✱0/4

Code generation49.6−20.4—/77✱0/2

Knowledge45.2−18.3—/138✱0/2

Factuality42.8−16.2—/0✱✱

§ 2 · Head to head

What it beats, and what beats it

The same models turn up scope after scope. Counted once: where both were placed, who finished higher. Bars run right for Nemotron 3 Super, left for the other.

GPT-4o Miniwon 9 · lost 0

GPT-4.1 Nanowon 8 · lost 1

NVIDIA Nemotron 3 Nano 4Bwon 7 · lost 1

Mistral Medium 3.5won 7 · lost 1

Granite 4.2 3Bwon 8 · lost 2

Command Awon 7 · lost 2

G9v3 3Bwon 6 · lost 2

Mistral Large 3won 6 · lost 2

Granite 4.2 8Bwon 7 · lost 3

Grok 4.3won 6 · lost 3

Gemini 3.5 Flash Litewon 6 · lost 3

GPT-5.4 Miniwon 6 · lost 3

K2 Horizon 0.9Bwon 5 · lost 3

Granite 4.2 30Bwon 7 · lost 5

GPT-5.4 Nanowon 5 · lost 4

Qwen3.6 Pluswon 5 · lost 4

DeepSeek V3won 5 · lost 4

Llama 4 Maverick Instructwon 5 · lost 4

G9v3 39A5Bwon 6 · lost 6

Kimi K2.6won 4 · lost 5

GPT-4.1 Miniwon 4 · lost 5

Qwen3.5 9Bwon 4 · lost 5

Gemini 3.1 Prowon 4 · lost 5

MiniMax M3won 5 · lost 8

Qwen3.5 4Bwon 3 · lost 5

GPT OSS 120Bwon 3 · lost 5

Gemma 4 12Bwon 3 · lost 5

Mimo V2.5 Prowon 3 · lost 5

Gemini 3.1 Flash Litewon 3 · lost 5

Claude Haiku 4.5won 3 · lost 6

GLM-4.6won 3 · lost 6

Qwen3.7 Maxwon 3 · lost 6

Gemini 2.5 Prowon 3 · lost 6

K2 Horizon 32Bwon 4 · lost 8

Qwen3.6 35B A3Bwon 4 · lost 9

Inklingwon 4 · lost 9

Nemotron 3 Ultrawon 4 · lost 9

K2 Horizon 3.7Bwon 3 · lost 7

K2 Horizon 7Bwon 3 · lost 7

DeepSeek R1won 2 · lost 6

Claude Opus 4won 2 · lost 6

Kimi K2.5won 2 · lost 6

Gemma 4 31Bwon 3 · lost 10

Kimi K2won 2 · lost 7

DeepSeek V3.2won 2 · lost 7

GPT-4.1won 2 · lost 7

Claude Sonnet 4.5won 2 · lost 7

Qwen3 235B A22B Instructwon 2 · lost 7

GPT-5 Miniwon 2 · lost 7

Claude Sonnet 4won 2 · lost 7

Gemini 3 Prowon 2 · lost 7

DeepSeek V4 Flashwon 2 · lost 8

Claude Opus 4.5won 2 · lost 8

O4 Miniwon 2 · lost 8

Qwen3.8 27Bwon 2 · lost 11

GLM-5.3 Flashwon 1 · lost 8

Muse Spark 1.1won 1 · lost 8

Gemini 3.8 Flashwon 1 · lost 8

Grok 4.5won 1 · lost 8

Grok 4.6won 1 · lost 8

Qwen3.8 Maxwon 1 · lost 8

Gemini 2.5 Flashwon 1 · lost 9

GPT-5.2won 1 · lost 9

DeepSeek V4 Prowon 1 · lost 9

DeepSeek V4.1 Flashwon 1 · lost 9

K2 Horizon MoVA 36B A4Bwon 1 · lost 12

Claude Opus 4.6won 0 · lost 8

Grok 3 Miniwon 0 · lost 8

Muse Spark 1.2won 0 · lost 8

GPT-5won 0 · lost 8

Claude Sonnet 4.6won 0 · lost 9

Gemini 3.6 Flashwon 0 · lost 9

Gemini 3.5 Flashwon 0 · lost 9

Claude Opus 4.8won 0 · lost 9

Claude Opus 4.7won 0 · lost 9

GPT-5.4won 0 · lost 9

Grok 4won 0 · lost 9

Gemini 3.7 Flashwon 0 · lost 9

GPT-5.5won 0 · lost 9

Claude Fable 5.1won 0 · lost 9

Claude Fable 5won 0 · lost 9

GLM-5.3won 0 · lost 10

O3won 0 · lost 10

Claude Opus 5won 0 · lost 10

Kimi K3won 0 · lost 10

GPT-5.6 Solwon 0 · lost 10

Muse Glimmer-30Bwon 0 · lost 12

GPT-5.6 Lunawon 0 · lost 13

GLM-5.2won 0 · lost 13

Claude Sonnet 5won 0 · lost 13

K2 Horizon 375B A23Bwon 0 · lost 13

GPT-5.6 Terrawon 0 · lost 13

§ 3 · Sources

Where the numbers come from

2 publications, 10 figures. Every one links to the page it was read from.

Artificial Analysis ↗1 measureread 2026-09-27

Artificial Analysis Intelligence Index 13

✱✱ IFM — K2 Horizon launch ↗the model’s own publisher9 measuresread 2026-09-27

tau3-Banking 10.3%Terminal-Bench 2.1 38.6%SciCode 36%Humanity's Last Exam (without tools) 20.8%GPQA Diamond 80%CritPt 3.1%AA-LCR 60.3%AA-Omniscience Accuracy 24.3%AA-Omniscience Non-Hallucination 13%

Badge

[![PublicAI Index](https://publicai.io/model-index/badge?model=nemotron-3-super)](https://publicai.io/model-index/m/nemotron-3-super)