Compare AI models

Specifications
Parameters
16B—
Context window
32k256k
Benchmarks
HumanEval
57.3%86.6%
Overview
CompanyDeepSeekMistral
Release dateMay 16 2024Jan 13 2025
AccessOpen WeightClosed
Model detailsView modelView model

Frequently asked questions

Codestral 25.01 leads DeepSeek V2 Lite on 1 of the 1 benchmark they both report (HumanEval). DeepSeek V2 Lite shipped 242 days before Codestral 25.01, so benchmark comparisons should account for the intervening progress.

Context windows are 32k (DeepSeek V2 Lite) vs 256k (Codestral 25.01). DeepSeek V2 Lite is open weight, while Codestral 25.01 is closed.

On HumanEval, Codestral 25.01 leads at 86.6% vs DeepSeek V2 Lite at 57.3%.