Compare / DeepSeek V4 Flash 0423 vs Kimi K2.6

DeepSeek V4 Flash 0423 vs Kimi K2.6

19 results from 3 sources that measured both models. Each row is in the source's own units; there is no overall winner.

DeepSeek V4 Flash 0423
DeepSeek · profile
Kimi K2.6
Moonshot AI · profile
Results compared
19
Shared sources
3
BenchmarkMetricDeepSeek V4 Flash 0423Kimi K2.6
Arena (formerly LMArena)reportedOverallArena rating 1,4391,456
Arena (formerly LMArena)reportedBusiness, management and financeArena rating 1,4391,455
Arena (formerly LMArena)reportedCreative writingArena rating 1,4081,433
Arena (formerly LMArena)reportedExpert promptsArena rating 1,4571,503
Arena (formerly LMArena)reportedInstruction followingArena rating 1,4281,456
Arena (formerly LMArena)reportedOverallArena rating 1,4361,461
Arena (formerly LMArena)reportedWriting, literature and languageArena rating 1,4161,441
Artificial AnalysisreportedArtificial Analysis Coding Indexindex score 69.161.8
Artificial AnalysisreportedArtificial Analysis Intelligence Indexindex score 34.327.0
Artificial AnalysisreportedGPQA Diamond% of questions 90.8%91.1%
Artificial AnalysisreportedHumanity's Last Exam% of questions 38.6%37.5%
Artificial AnalysisreportedLong-context reasoning (AA-LCR)% of questions 79.7%81.0%
Artificial AnalysisreportedSciCode% of problems 50.3%51.5%
Artificial AnalysisreportedTerminal-Bench 2.1% of tasks 78.7%65.9%
Artificial AnalysisreportedTerminal-Bench 4.0% of tasks 12.1%0.5%
Artificial AnalysisreportedΤ-bench banking% of tasks 39.4%23.3%
UGI LeaderboardreportedRequested-length error% off the requested word count 12.0%18.0%
UGI LeaderboardreportedStyle adherencescore from 0 to 1 0.360.35
UGI LeaderboardreportedWriting scorescore out of 100 54.664.4

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.