Compare AI models

Specifications
Parameters
671B671B
Context window
128k128k
API pricing
Cheapest input
$0.256DeepInfra$0.27SiliconFlow
Cheapest output
$0.712DeepInfra$0.41SiliconFlow
$103StreamLake$74.50SiliconFlow · fp8
Benchmarks
SWE-Bench Verified
42%67.8%
LiveCodeBench
40.5%74.1%
GPQA Diamond
59.1%79.9%
Overview
CompanyDeepSeekDeepSeek
Release dateDec 26 2024Sep 29 2025
AccessOpen WeightOpen Weight
Model detailsView modelView model

Frequently asked questions

DeepSeek V3.2 Exp leads DeepSeek V3 on 3 of the 3 benchmarks they both report (SWE-Bench Verified, LiveCodeBench, GPQA Diamond). DeepSeek V3 shipped 277 days before DeepSeek V3.2 Exp, so benchmark comparisons should account for the intervening progress.

DeepSeek V3 has 671B parameters, while DeepSeek V3.2 Exp has 671B. Context windows are 128k (DeepSeek V3) vs 128k (DeepSeek V3.2 Exp).

On SWE-Bench Verified, DeepSeek V3.2 Exp leads at 67.8% vs DeepSeek V3 at 42%. On LiveCodeBench, DeepSeek V3.2 Exp leads at 74.1% vs DeepSeek V3 at 40.5%. On GPQA Diamond, DeepSeek V3.2 Exp leads at 79.9% vs DeepSeek V3 at 59.1%.