NVIDIA logo

Llama-3.3 Nemotron Super 49B v1

NVIDIA

Llama 3.3 Nemotron Super 49B v1 is a high-performance language model by NVIDIA, built on the Llama 3.3 architecture and enhanced with NVIDIA's Nemotron training methodology. With 49 billion parameters, it delivers exceptional performance in reasoning, coding, and instruction following, optimized for enterprise and research applications.

Key Specifications

Parameters
49.9B
Context
-
Release Date
March 18, 2025
Average Score
81.0%

Timeline

Key dates in the model's history
Announcement
March 18, 2025
Last Update
July 19, 2025
Today
March 25, 2026

Technical Specifications

Parameters
49.9B
Training Tokens
-
Knowledge Cutoff
December 31, 2023
Family
-
Capabilities
MultimodalZeroEval

Benchmark Results

Model performance metrics across various tests and benchmarks

Programming

Programming skills tests
MBPP
Pass@1, Reasoning On — Pass@1 Pass@1Self-reported
91.3%

Reasoning

Logical reasoning and analysis
GPQA
Pass@1, Pass@1 — Pass@1 "" LLMSelf-reported
66.7%

Other Tests

Specialized benchmarks
AIME 2025
Pass@1, Pass@1Self-reported
58.4%
Arena Hard
Score, Reasoning Off AI: Standard Score: — evaluation {/} evaluation STEM, "" : :Self-reported
88.3%
BFCL v2
**Score, Reasoning On** [Score, Reasoning Optional], Anthropic 0, ("3/8" "0.375")Self-reported
73.7%
MATH-500
Pass@1, Reasoning On "5 ?", "78.5 ", πr²Self-reported
96.6%
MT-Bench
Score, Reasoning On (Score)Self-reported
91.7%

License & Metadata

License
llama_3_1_community_license
Announcement Date
March 18, 2025
Last Updated
July 19, 2025

Similar Models

All Models

Recommendations are based on similarity of characteristics: developer organization, multimodality, parameter size, and benchmark performance. Choose a model to compare or go to the full catalog to browse all available AI models.