Compare / Kimi K2.5 vs Kimi K3

Kimi K2.5 vs Kimi K3

16 results from 4 sources that measured both models. Each row is in the source's own units; there is no overall winner.

Kimi K2.5
Moonshot AI · profile
Kimi K3
Moonshot AI · profile
Results compared
16
Shared sources
4
BenchmarkMetricKimi K2.5Kimi K3
Arena (formerly LMArena)reportedOverallArena rating 1,4451,472
Arena (formerly LMArena)reportedBusiness, management and financeArena rating 1,4411,489
Arena (formerly LMArena)reportedCreative writingArena rating 1,4231,458
Arena (formerly LMArena)reportedExpert promptsArena rating 1,4831,533
Arena (formerly LMArena)reportedInstruction followingArena rating 1,4391,486
Arena (formerly LMArena)reportedOverallArena rating 1,4501,488
Arena (formerly LMArena)reportedWriting, literature and languageArena rating 1,4281,475
Artificial AnalysisreportedArtificial Analysis Coding Indexindex score 46.876.2
Artificial AnalysisreportedArtificial Analysis Intelligence Indexindex score 23.543.6
Artificial AnalysisreportedGPQA Diamond% of questions 87.9%93.5%
Artificial AnalysisreportedHumanity's Last Exam% of questions 30.7%46.9%
Artificial AnalysisreportedLong-context reasoning (AA-LCR)% of questions 78.0%88.7%
Artificial AnalysisreportedTerminal-Bench 2.1% of tasks 45.7%85.0%
Artificial AnalysisreportedΤ-bench banking% of tasks 14.2%46.0%
SimpleQA Verified (Epoch AI)reportedCorrect answers% of questions 34.3%50.6%
Vending-Bench 2reportedMoney after a yearUS dollars $1,198.46$5,165.03

Bold green marks the better value on that metric. Where a source reports ranges that overlap, the difference may not be meaningful; see the benchmark page for ranges.