Compare AI models

Specifications
Parameters
—671B
Context window
100k128k
Benchmarks
MMLU
78.5%90.8%
Overview
CompanyAnthropicDeepSeek
Release dateJul 11 2023Jan 20 2025
AccessClosedOpen Weight
Model detailsView modelView model

Frequently asked questions

DeepSeek R1 leads Claude 2 on 1 of the 1 benchmark they both report (MMLU). Claude 2 shipped 559 days before DeepSeek R1, so benchmark comparisons should account for the intervening progress.

Context windows are 100k (Claude 2) vs 128k (DeepSeek R1). Claude 2 is closed, while DeepSeek R1 is open weight.

On MMLU, DeepSeek R1 leads at 90.8% vs Claude 2 at 78.5%.