Compare AI models

Specifications
Parameters
7B/13B/70B7B
Context window
4k32k
Benchmarks
HumanEval
29.9%38.4%
MATH
13.5%13.2%
Overview
CompanyMetaMistral
Release dateJul 18 2023May 22 2024
AccessOpen WeightOpen Weight
Model detailsView modelView model

Frequently asked questions

LLaMA 2 and Mistral 7B v0.3 are evenly matched across the 2 benchmarks they both report (HumanEval, MATH). LLaMA 2 shipped 309 days before Mistral 7B v0.3, so benchmark comparisons should account for the intervening progress.

LLaMA 2 has 7B/13B/70B parameters, while Mistral 7B v0.3 has 7B. Context windows are 4k (LLaMA 2) vs 32k (Mistral 7B v0.3).

On HumanEval, Mistral 7B v0.3 leads at 38.4% vs LLaMA 2 at 29.9%. On MATH, LLaMA 2 leads at 13.5% vs Mistral 7B v0.3 at 13.2%.