Compare AI models

Specifications
Parameters
1.5B/7B/8B/14B/32B/70B—
Context window
—128k
Benchmarks
MATH-500
94.5%85.5%
AIME 2024
70%44.6%
GPQA Diamond
65.2%73.3%
Overview
CompanyDeepSeekOpenAI
Release dateJan 20 2025Sep 12 2024
AccessOpen WeightClosed
Model detailsView modelView model

Frequently asked questions

DeepSeek R1 Distill leads o1-preview on 2 of the 3 benchmarks they both report (MATH-500, AIME 2024, GPQA Diamond). o1-preview shipped 130 days before DeepSeek R1 Distill, so benchmark comparisons should account for the intervening progress.

DeepSeek R1 Distill is open weight, while o1-preview is closed.

On MATH-500, DeepSeek R1 Distill leads at 94.5% vs o1-preview at 85.5%. On AIME 2024, DeepSeek R1 Distill leads at 70% vs o1-preview at 44.6%. On GPQA Diamond, o1-preview leads at 73.3% vs DeepSeek R1 Distill at 65.2%.