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

Kimi K2 Thinking vs DeepSeek V3.2 Speciale

Côte à côte. Chaque métrique. Chaque benchmark.

moonshotai
53.3
score moyen
1/6
benchmarks
DeepSeek
78.2
score moyen
5/6
benchmarks
TypeKimi K2 ThinkingDeepSeek V3.2 Speciale
Providermoonshotai logomoonshotaiDeepSeek logoDeepSeek
score moyen53.378.2
Prix d'entrée$0.60$0.40
Prix de sortie$2.50$1.20
Fenêtre de contexte262K tokens (~131 books)164K tokens (~82 books)
Sorti le2025-11-062025-12-01
Code source ouvertOpen SourceOpen Source

6 benchmarks · Kimi K2 Thinking: 1, DeepSeek V3.2 Speciale: 5

BenchmarkCatégorieKimi K2 ThinkingDeepSeek V3.2 Speciale
OpenCompass — AIME2025math94.196.0
OpenCompass — GPQA-Diamondknowledge82.786.7
OpenCompass — HLEknowledge21.328.6
OpenCompass — IFEvallanguage92.491.7
OpenCompass — LiveCodeBenchV6coding77.180.9
OpenCompass — MMLU-Proknowledge84.385.5