Compare AI models

Specifications
Parameters
141B51B
Context window
64k128k
API pricing
Cheapest input
$2.00Mistral—
Cheapest output
$6.00Mistral—
$700Mistral · zdr—
Benchmarks
ARC-Challenge
91.3%69.2%
HellaSwag
88.5%85.58%
WinoGrande
84.7%84.53%
MMLU
77.75%80.2%
GSM8K
78.6%91.43%
Overview
CompanyMistralNVIDIA
Release dateApr 10 2024Sep 23 2024
AccessOpen WeightOpen Weight
Model detailsView modelView model

Frequently asked questions

Mixtral 8×22B leads Llama 3.1 Nemotron 51B on 3 of the 5 benchmarks they both report. Mixtral 8×22B shipped 166 days before Llama 3.1 Nemotron 51B, so benchmark comparisons should account for the intervening progress.

Mixtral 8×22B has 141B parameters, while Llama 3.1 Nemotron 51B has 51B. Context windows are 64k (Mixtral 8×22B) vs 128k (Llama 3.1 Nemotron 51B).

On ARC-Challenge, Mixtral 8×22B leads at 91.3% vs Llama 3.1 Nemotron 51B at 69.2%. On HellaSwag, Mixtral 8×22B leads at 88.5% vs Llama 3.1 Nemotron 51B at 85.58%. On WinoGrande, Mixtral 8×22B leads at 84.7% vs Llama 3.1 Nemotron 51B at 84.53%. On MMLU, Llama 3.1 Nemotron 51B leads at 80.2% vs Mixtral 8×22B at 77.75%. On GSM8K, Llama 3.1 Nemotron 51B leads at 91.43% vs Mixtral 8×22B at 78.6%.