Compare AI models

Specifications
Parameters
16B—
Context window
32k32k
Benchmarks
HumanEval
57.3%45.1%
MMLU
55.7%81.2%
GSM8K
72%81%
Overview
CompanyDeepSeekMistral
Release dateMay 16 2024Feb 26 2024
AccessOpen WeightClosed
Model detailsView modelView model

Frequently asked questions

Mistral Large leads DeepSeek V2 Lite on 2 of the 3 benchmarks they both report (HumanEval, MMLU, GSM8K). Mistral Large shipped 80 days before DeepSeek V2 Lite, so benchmark comparisons should account for the intervening progress.

Context windows are 32k (DeepSeek V2 Lite) vs 32k (Mistral Large). DeepSeek V2 Lite is open weight, while Mistral Large is closed.

On HumanEval, DeepSeek V2 Lite leads at 57.3% vs Mistral Large at 45.1%. On MMLU, Mistral Large leads at 81.2% vs DeepSeek V2 Lite at 55.7%. On GSM8K, Mistral Large leads at 81% vs DeepSeek V2 Lite at 72%.