Compare AI models

Specifications
Parameters
7B14B
Context window
32k2k
Benchmarks
HumanEval
38.4%32.3%
MATH
13.2%24.8%
Overview
CompanyMistralQwen
Release dateMay 22 2024Sep 25 2023
AccessOpen WeightOpen Weight
Model detailsView modelView model

Frequently asked questions

Mistral 7B v0.3 and Qwen-14B are evenly matched across the 2 benchmarks they both report (HumanEval, MATH). Qwen-14B shipped 240 days before Mistral 7B v0.3, so benchmark comparisons should account for the intervening progress.

Mistral 7B v0.3 has 7B parameters, while Qwen-14B has 14B. Context windows are 32k (Mistral 7B v0.3) vs 2k (Qwen-14B).

On HumanEval, Mistral 7B v0.3 leads at 38.4% vs Qwen-14B at 32.3%. On MATH, Qwen-14B leads at 24.8% vs Mistral 7B v0.3 at 13.2%.