Compare AI models

Specifications
Parameters
671B—
Context window
128k8k
API pricing
Cheapest input
$0.256DeepInfra$30.00Azure
Cheapest output
$0.712DeepInfra$60.00Azure
$103StreamLake$9,000Azure
Benchmarks
MMLU
88.5%86.4%
Overview
CompanyDeepSeekOpenAI
Release dateDec 26 2024Mar 14 2023
AccessOpen WeightClosed
Model detailsView modelView model

Frequently asked questions

DeepSeek V3 leads GPT-4 on 1 of the 1 benchmark they both report (MMLU). GPT-4 shipped 653 days before DeepSeek V3, so benchmark comparisons should account for the intervening progress.

Context windows are 128k (DeepSeek V3) vs 8k (GPT-4). DeepSeek V3 is open weight, while GPT-4 is closed.

On MMLU, DeepSeek V3 leads at 88.5% vs GPT-4 at 86.4%.