Beta
Home/Comparer/DeepSeek V3.2 Speciale vs Kimi K2.5

DeepSeek V3.2 Speciale vs Kimi K2.5

Côte à côte. Chaque métrique. Chaque benchmark.

DeepSeek
78.2
score moyen
2/9
benchmarks
moonshotai
52.0
score moyen
6/9
benchmarks
TypeDeepSeek V3.2 SpecialeKimi K2.5
ProviderDeepSeek logoDeepSeekmoonshotai logomoonshotai
score moyen78.252.0
Prix d'entrée$0.40$0.38
Prix de sortie$1.20$1.72
Fenêtre de contexte164K tokens (~82 books)262K tokens (~131 books)
Sorti le2025-12-012026-01-27
Code source ouvertOpen SourceOpen Source

9 benchmarks · DeepSeek V3.2 Speciale: 2, Kimi K2.5: 6

BenchmarkCatégorieDeepSeek V3.2 SpecialeKimi K2.5
Artificial Analysis — Agentic Indexspeed0.058.9
Artificial Analysis — Coding Indexspeed37.939.5
Artificial Analysis — Quality Indexspeed29.446.8
OpenCompass — AIME2025math96.091.9
OpenCompass — GPQA-Diamondknowledge86.788.1
OpenCompass — HLEknowledge28.628.6
OpenCompass — IFEvallanguage91.793.9
OpenCompass — LiveCodeBenchV6coding80.980.6
OpenCompass — MMLU-Proknowledge85.586.2