Compare AI models

Specifications
Parameters
9B/27B51B
Context window
8k128k
API pricing
Cheapest input
$0.65NextBit—
Cheapest output
$0.65NextBit—
$163NextBit · int4—
Benchmarks
ARC-Challenge
71.4%69.2%
HellaSwag
86.4%85.58%
MMLU
75.2%80.2%
GSM8K
74%91.43%
Overview
CompanyGoogleNVIDIA
Release dateJun 27 2024Sep 23 2024
AccessOpen WeightOpen Weight
Model detailsView modelView model

Frequently asked questions

Gemma 2 and Llama 3.1 Nemotron 51B are evenly matched across the 4 benchmarks they both report (ARC-Challenge, HellaSwag, MMLU, GSM8K). Gemma 2 shipped 88 days before Llama 3.1 Nemotron 51B, so benchmark comparisons should account for the intervening progress.

Gemma 2 has 9B/27B parameters, while Llama 3.1 Nemotron 51B has 51B. Context windows are 8k (Gemma 2) vs 128k (Llama 3.1 Nemotron 51B).

On ARC-Challenge, Gemma 2 leads at 71.4% vs Llama 3.1 Nemotron 51B at 69.2%. On HellaSwag, Gemma 2 leads at 86.4% vs Llama 3.1 Nemotron 51B at 85.58%. On MMLU, Llama 3.1 Nemotron 51B leads at 80.2% vs Gemma 2 at 75.2%. On GSM8K, Llama 3.1 Nemotron 51B leads at 91.43% vs Gemma 2 at 74%.