Compare AI models

Specifications
Parameters
671B—
Context window
128k128k
Benchmarks
SWE-Bench Verified
49.2%38%
MATH-500
97.3%87.1%
DROP
92.283.4
AIME 2024
79.8%36.7%
GPQA Diamond
71.5%69.5%
SimpleQA
30.1%62.5%
MMLU
90.8%90.8%
Overview
CompanyDeepSeekOpenAI
Release dateJan 20 2025Feb 27 2025
AccessOpen WeightClosed
Model detailsView modelView model

Frequently asked questions

DeepSeek R1 leads GPT-4.5 on 5 of the 7 benchmarks they both report. DeepSeek R1 shipped 38 days before GPT-4.5, so benchmark comparisons should account for the intervening progress.

Context windows are 128k (DeepSeek R1) vs 128k (GPT-4.5). DeepSeek R1 is open weight, while GPT-4.5 is closed.

On SWE-Bench Verified, DeepSeek R1 leads at 49.2% vs GPT-4.5 at 38%. On MATH-500, DeepSeek R1 leads at 97.3% vs GPT-4.5 at 87.1%. On DROP, DeepSeek R1 leads at 92.2 vs GPT-4.5 at 83.4. On AIME 2024, DeepSeek R1 leads at 79.8% vs GPT-4.5 at 36.7%. On GPQA Diamond, DeepSeek R1 leads at 71.5% vs GPT-4.5 at 69.5%. On SimpleQA, GPT-4.5 leads at 62.5% vs DeepSeek R1 at 30.1%. On MMLU, both models score 90.8%.