Compare AI models

Specifications
Parameters
671B1T
Context window
128k256k
API pricing
Cheapest input
—$0.49AtlasCloud
Cheapest output
—$2.50AtlasCloud
—$223AtlasCloud · int4
Benchmarks
LiveCodeBench
88.7%85%
GPQA Diamond
85.7%87.6%
Overview
CompanyDeepSeekMoonshot AI
Release dateDec 1 2025Jan 27 2026
AccessOpen WeightOpen Weight
Model detailsView modelView model

Frequently asked questions

DeepSeek-V3.2-Speciale and Kimi K2.5 are evenly matched across the 2 benchmarks they both report (LiveCodeBench, GPQA Diamond). DeepSeek-V3.2-Speciale shipped 57 days before Kimi K2.5, so benchmark comparisons should account for the intervening progress.

DeepSeek-V3.2-Speciale has 671B parameters, while Kimi K2.5 has 1T. Context windows are 128k (DeepSeek-V3.2-Speciale) vs 256k (Kimi K2.5).

On LiveCodeBench, DeepSeek-V3.2-Speciale leads at 88.7% vs Kimi K2.5 at 85%. On GPQA Diamond, Kimi K2.5 leads at 87.6% vs DeepSeek-V3.2-Speciale at 85.7%.