Compare AI models

Specifications
Parameters
236B1B/3B/11B/90B
Context window
128k128k
API pricing
Cheapest input
—$0.027Cloudflare
Cheapest output
—$0.201Cloudflare
—$15.45Cloudflare
Benchmarks
MATH
71.02%68%
Overview
CompanyDeepSeekMeta
Release dateJun 28 2024Sep 25 2024
AccessOpen WeightOpen Weight
Model detailsView modelView model

Frequently asked questions

DeepSeek V2-0628 leads LLaMA 3.2 on 1 of the 1 benchmark they both report (MATH). DeepSeek V2-0628 shipped 89 days before LLaMA 3.2, so benchmark comparisons should account for the intervening progress.

DeepSeek V2-0628 has 236B parameters, while LLaMA 3.2 has 1B/3B/11B/90B. Context windows are 128k (DeepSeek V2-0628) vs 128k (LLaMA 3.2).

On MATH, DeepSeek V2-0628 leads at 71.02% vs LLaMA 3.2 at 68%.