Models/Llama-3_3-Nemotron-Super-49B-v1

MDL

Llama-3_3-Nemotron-Super-49B-v1

nvidia

Hugging Face ↗

Verified against Artificial Analysis · Jul 13, 2026

Parameters

49B

Price /M

$0 / $0

in / out

Context

—

Avg score

39.8

11 benchmarks

AA Intelligence

12.2

index

Weights

Open

Released

Mar 18, 2025

About

Llama-3_3-Nemotron-Super-49B-v1 is a 49B-parameter open-weights model in the Llama family from nvidia. Benchmarked on 11 evals, averaging 39.8.

othertext-generation

Benchmark scores · 11

avg 39.8 — higher is better

MATHartificial-analysis · 2026-07

95.9

MMLU-Proartificial-analysis · 2026-07

78.5

GPQA Diamondartificial-analysis · 2026-07

64.3

AIMEartificial-analysis · 2026-07

54.7

IFBenchartificial-analysis · 2026-07

38.1

SciCodeartificial-analysis · 2026-07

28.2

LiveCodeBenchartificial-analysis · 2026-07

27.7

τ²-benchartificial-analysis · 2026-07

26.9

AA Long-Context Reasoningartificial-analysis · 2026-07

Humanity's Last Examartificial-analysis · 2026-07

6.5

Terminal-Bench Hardartificial-analysis · 2026-07

The research behind this model

all papers mentioning it →

Puzzle: Distillation-Based NAS for Inference-Optimized LLMs Llama-Nemotron: Efficient Reasoning Models Reward-aware Preference Optimization: A Unified Mathematical Framework for Model Alignment

Details

Publishernvidia

FamilyLlama

Parameters49B

Licenseother

WeightsOpen

ReleasedMar 18, 2025

History

full ledger →

Capability over time

Append-only ledger — every observed change to this model's numbers.

API providers

Artificial Analysis$0/M