Compare AI models

Specifications
Parameters
—671B
Context window
200k128k
API pricing
Cheapest input
—$0.256DeepInfra
Cheapest output
—$0.712DeepInfra
—$103StreamLake
Benchmarks
SWE-Bench Verified
40.6%42%
GPQA Diamond
41.6%59.1%
MMLU
80.9%88.5%
Overview
CompanyAnthropicDeepSeek
Release dateOct 22 2024Dec 26 2024
AccessClosedOpen Weight
Model detailsView modelView model

Frequently asked questions

DeepSeek V3 leads Claude 3.5 Haiku on 3 of the 3 benchmarks they both report (SWE-Bench Verified, GPQA Diamond, MMLU). Claude 3.5 Haiku shipped 65 days before DeepSeek V3, so benchmark comparisons should account for the intervening progress.

Context windows are 200k (Claude 3.5 Haiku) vs 128k (DeepSeek V3). Claude 3.5 Haiku is closed, while DeepSeek V3 is open weight.

On SWE-Bench Verified, DeepSeek V3 leads at 42% vs Claude 3.5 Haiku at 40.6%. On GPQA Diamond, DeepSeek V3 leads at 59.1% vs Claude 3.5 Haiku at 41.6%. On MMLU, DeepSeek V3 leads at 88.5% vs Claude 3.5 Haiku at 80.9%.