Compare AI models

Specifications
Parameters
671B24B
Context window
128k128k
API pricing
Cheapest input
—$0.351Cloudflare
Cheapest output
—$0.555Cloudflare
—$97.95Cloudflare
Benchmarks
GPQA Diamond
71.5%45.96%
SimpleQA
30.1%10.43%
MMLU-Pro
84%66.76%
MMLU
90.8%80.62%
Overview
CompanyDeepSeekMistral
Release dateJan 20 2025Mar 17 2025
AccessOpen WeightOpen Weight
Model detailsView modelView model

Frequently asked questions

DeepSeek R1 leads Mistral Small 3.1 on 4 of the 4 benchmarks they both report (GPQA Diamond, SimpleQA, MMLU-Pro, MMLU). DeepSeek R1 shipped 56 days before Mistral Small 3.1, so benchmark comparisons should account for the intervening progress.

DeepSeek R1 has 671B parameters, while Mistral Small 3.1 has 24B. Context windows are 128k (DeepSeek R1) vs 128k (Mistral Small 3.1).

On GPQA Diamond, DeepSeek R1 leads at 71.5% vs Mistral Small 3.1 at 45.96%. On SimpleQA, DeepSeek R1 leads at 30.1% vs Mistral Small 3.1 at 10.43%. On MMLU-Pro, DeepSeek R1 leads at 84% vs Mistral Small 3.1 at 66.76%. On MMLU, DeepSeek R1 leads at 90.8% vs Mistral Small 3.1 at 80.62%.