Compare AI models

Specifications
Parameters
671B—
Context window
128k128k
Benchmarks
SWE-Bench Verified
49.2%41.4%
MATH-500
97.3%85.5%
DROP
92.274.8
AIME 2024
79.8%44.6%
GPQA Diamond
71.5%73.3%
MMLU
90.8%90.8%
Overview
CompanyDeepSeekOpenAI
Release dateJan 20 2025Sep 12 2024
AccessOpen WeightClosed
Model detailsView modelView model

Frequently asked questions

DeepSeek R1 leads o1-preview on 4 of the 6 benchmarks they both report. o1-preview shipped 130 days before DeepSeek R1, so benchmark comparisons should account for the intervening progress.

Context windows are 128k (DeepSeek R1) vs 128k (o1-preview). DeepSeek R1 is open weight, while o1-preview is closed.

On SWE-Bench Verified, DeepSeek R1 leads at 49.2% vs o1-preview at 41.4%. On MATH-500, DeepSeek R1 leads at 97.3% vs o1-preview at 85.5%. On DROP, DeepSeek R1 leads at 92.2 vs o1-preview at 74.8. On AIME 2024, DeepSeek R1 leads at 79.8% vs o1-preview at 44.6%. On GPQA Diamond, o1-preview leads at 73.3% vs DeepSeek R1 at 71.5%. On MMLU, both models score 90.8%.