Compare AI models

Specifications
Parameters
671B8B/70B/405B
Context window
128k128k
API pricing
Cheapest input
—$0.02DeepInfra
Cheapest output
—$0.04DeepInfra
—$6.00DeepInfra · fp8
Benchmarks
GPQA Diamond
71.5%49%
MMLU-Pro
84%73.3%
MMLU
90.8%87.3%
Overview
CompanyDeepSeekMeta
Release dateJan 20 2025Jul 23 2024
AccessOpen WeightOpen Weight
Model detailsView modelView model

Frequently asked questions

DeepSeek R1 leads LLaMA 3.1 on 3 of the 3 benchmarks they both report (GPQA Diamond, MMLU-Pro, MMLU). LLaMA 3.1 shipped 181 days before DeepSeek R1, so benchmark comparisons should account for the intervening progress.

DeepSeek R1 has 671B parameters, while LLaMA 3.1 has 8B/70B/405B. Context windows are 128k (DeepSeek R1) vs 128k (LLaMA 3.1).

On GPQA Diamond, DeepSeek R1 leads at 71.5% vs LLaMA 3.1 at 49%. On MMLU-Pro, DeepSeek R1 leads at 84% vs LLaMA 3.1 at 73.3%. On MMLU, DeepSeek R1 leads at 90.8% vs LLaMA 3.1 at 87.3%.