Compare AI models

Specifications
Parameters
671B—
Context window
128k128k
Benchmarks
DROP
92.283.2
GPQA Diamond
71.5%42.5%
MMLU
90.8%84.7%
Overview
CompanyDeepSeekOpenAI
Release dateJan 20 2025Nov 6 2023
AccessOpen WeightClosed
Model detailsView modelView model

Frequently asked questions

DeepSeek R1 leads GPT-4 Turbo on 3 of the 3 benchmarks they both report (DROP, GPQA Diamond, MMLU). GPT-4 Turbo shipped 441 days before DeepSeek R1, so benchmark comparisons should account for the intervening progress.

Context windows are 128k (DeepSeek R1) vs 128k (GPT-4 Turbo). DeepSeek R1 is open weight, while GPT-4 Turbo is closed.

On DROP, DeepSeek R1 leads at 92.2 vs GPT-4 Turbo at 83.2. On GPQA Diamond, DeepSeek R1 leads at 71.5% vs GPT-4 Turbo at 42.5%. On MMLU, DeepSeek R1 leads at 90.8% vs GPT-4 Turbo at 84.7%.