Compare AI models

Specifications
Parameters
16B24B
Context window
32k128k
API pricing
Cheapest input
—$0.351Cloudflare
Cheapest output
—$0.555Cloudflare
—$97.95Cloudflare
Benchmarks
HumanEval
57.3%88.41%
MATH
27.9%69.3%
MMLU
55.7%80.62%
Overview
CompanyDeepSeekMistral
Release dateMay 16 2024Mar 17 2025
AccessOpen WeightOpen Weight
Model detailsView modelView model

Frequently asked questions

Mistral Small 3.1 leads DeepSeek V2 Lite on 3 of the 3 benchmarks they both report (HumanEval, MATH, MMLU). DeepSeek V2 Lite shipped 305 days before Mistral Small 3.1, so benchmark comparisons should account for the intervening progress.

DeepSeek V2 Lite has 16B parameters, while Mistral Small 3.1 has 24B. Context windows are 32k (DeepSeek V2 Lite) vs 128k (Mistral Small 3.1).

On HumanEval, Mistral Small 3.1 leads at 88.41% vs DeepSeek V2 Lite at 57.3%. On MATH, Mistral Small 3.1 leads at 69.3% vs DeepSeek V2 Lite at 27.9%. On MMLU, Mistral Small 3.1 leads at 80.62% vs DeepSeek V2 Lite at 55.7%.