Compare AI models

Specifications
Parameters
16B22B
Context window
32k32k
Benchmarks
HumanEval
57.3%81.1%
Overview
CompanyDeepSeekMistral
Release dateMay 16 2024May 29 2024
AccessOpen WeightOpen Weight
Model detailsView modelView model

Frequently asked questions

Codestral 22B leads DeepSeek V2 Lite on 1 of the 1 benchmark they both report (HumanEval). DeepSeek V2 Lite shipped 13 days before Codestral 22B, so benchmark comparisons should account for the intervening progress.

DeepSeek V2 Lite has 16B parameters, while Codestral 22B has 22B. Context windows are 32k (DeepSeek V2 Lite) vs 32k (Codestral 22B).

On HumanEval, Codestral 22B leads at 81.1% vs DeepSeek V2 Lite at 57.3%.