Llama Nemotron Super 49B v1.5Open Weight

Released

Llama Nemotron Super 49B v1.5 is an AI model released by NVIDIA on Friday, Jul 25 2025, 66 days after Llama Nemotron Nano 4B v1.1. It is an open-weight model — the trained weights are available to download and run. It is a 49B parameterThe learned values that shape how a model responds. B means billion and T means trillion. For mixture-of-experts models, the total count is larger than the number active for each token. Parameter count alone does not measure quality. model with a 128k token context windowHow much information the model can work with in one request, measured in tokens — pieces of words or other data. This includes your question, attachments, conversation and reply. Limits can vary by model variant and provider.. Benchmark results (shown below) cover LiveCodeBench, Humanity's Last Exam, MATH-500, AIME 2024, AIME 2025, GPQA Diamond, and 1 more.

Benchmarks

Coding

LiveCodeBenchCompetitive coding — Coding problems published so recently the AI can't have seen them in training — a contamination-free test of raw programming skill. Higher is better.
73.58%
#32 of 86

Reasoning & science

Humanity's Last ExamMultidisciplinary reasoning — accuracy on the 2,158 text-only questions without tools
7.64%text only, no tools
#25 of 27
MATH-500Reasoning — accuracy on the 500-problem MATH subset
97.4%
#10 of 28
AIME 2024Reasoning — single-answer accuracy on the 2024 invitational mathematics exam without tools
87.5%
#9 of 30
AIME 2025Reasoning — single-answer accuracy on the 2025 invitational mathematics exam without tools
82.71%
#30 of 47
GPQA DiamondScience — Graduate-level science questions in biology, physics, and chemistry — hard enough that subject-matter PhDs score around 65%. Higher is better.
71.97%
#94 of 145

Knowledge work

MMLU-ProKnowledge — accuracy on the ten-option MMLU-Pro benchmark
79.53%
#30 of 58

Compare Llama Nemotron Super 49B v1.5 with

Llama Nemotron Super 49B v1.5

About

NVIDIA released Llama Nemotron Super 49B v1.5 in July 2025 as an update to its 49B Super tier. NVIDIA derived that tier from Meta’s Llama 3.3 70B Instruct through architecture search and distillation. It improved reasoning, coding, instruction following and function calling within the same size tier.

NVIDIA published the weights and the evaluation recipes. With reasoning enabled, the card reported 71.97% on GPQA and 97.4% on MATH-500.

Frequently asked questions

Llama Nemotron Super 49B v1.5 was released by NVIDIA on Friday, Jul 25 2025.

All NVIDIA releases

20 tracked