Compare AI models

Specifications
Parameters
671B14.3B
Context window
128k32k
API pricing
Cheapest input
$0.256DeepInfra—
Cheapest output
$0.712DeepInfra—
$103StreamLake—
Benchmarks
MMLU
88.5%62.5%
Overview
CompanyDeepSeekQwen
Release dateDec 26 2024Mar 28 2024
AccessOpen WeightOpen Weight
Model detailsView modelView model

Frequently asked questions

DeepSeek V3 leads Qwen1.5-MoE on 1 of the 1 benchmark they both report (MMLU). Qwen1.5-MoE shipped 273 days before DeepSeek V3, so benchmark comparisons should account for the intervening progress.

DeepSeek V3 has 671B parameters, while Qwen1.5-MoE has 14.3B. Context windows are 128k (DeepSeek V3) vs 32k (Qwen1.5-MoE).

On MMLU, DeepSeek V3 leads at 88.5% vs Qwen1.5-MoE at 62.5%.