Compare AI models

Specifications
Parameters
236B7B
Context window
128k256k
Benchmarks
HumanEval
84.76%75%
Overview
CompanyDeepSeekMistral
Release dateJun 28 2024Jul 16 2024
AccessOpen WeightOpen Weight
Model detailsView modelView model

Frequently asked questions

DeepSeek V2-0628 leads Codestral Mamba 7B on 1 of the 1 benchmark they both report (HumanEval). DeepSeek V2-0628 shipped 18 days before Codestral Mamba 7B, so benchmark comparisons should account for the intervening progress.

DeepSeek V2-0628 has 236B parameters, while Codestral Mamba 7B has 7B. Context windows are 128k (DeepSeek V2-0628) vs 256k (Codestral Mamba 7B).

On HumanEval, DeepSeek V2-0628 leads at 84.76% vs Codestral Mamba 7B at 75%.