Compare AI models

Specifications
Parameters
16B7B
Context window
32k32k
Benchmarks
MATH
27.9%56.6%
GSM8K
72%77.1%
Overview
CompanyDeepSeekMistral
Release dateMay 16 2024Jul 16 2024
AccessOpen WeightOpen Weight
Model detailsView modelView model

Frequently asked questions

Mathstral 7B leads DeepSeek V2 Lite on 2 of the 2 benchmarks they both report (MATH, GSM8K). DeepSeek V2 Lite shipped 61 days before Mathstral 7B, so benchmark comparisons should account for the intervening progress.

DeepSeek V2 Lite has 16B parameters, while Mathstral 7B has 7B. Context windows are 32k (DeepSeek V2 Lite) vs 32k (Mathstral 7B).

On MATH, Mathstral 7B leads at 56.6% vs DeepSeek V2 Lite at 27.9%. On GSM8K, Mathstral 7B leads at 77.1% vs DeepSeek V2 Lite at 72%.