0

Kimi K2 Thinking

Kimi K2 Thinking is an AI model from Kimi.

Publisher
Kimi
Family
Kimi
Context
262K
Input
Output
Released
6 Nov 2025

Cite

Notes

Only stored in your browser.

Attribution

Benchmark scores
frontiermathAALiveBenchSWE-benchVals AI
Attribution policy →
Context
262K
Input
$0.60
/1M tok
Output
$2.50
/1M tok
Blended
$1.07
#322/ 458/1M · 3:1
Speed
130
#77/ 458tok/s
Cache read
$0.15
0.25x input

Price & speed via Artificial Analysis · ranked across tracked models · caching via OpenRouter

Reported on 20 evals across 8 domains - best 94.7% on AIME 2025: Problems from the American Invitational Mathematics Examination · #6 of 207

Reported eval scores

20
LiveCodeBench85.3
SWE-bench Verified63.4
AIME 2025: Problems from the American Invitational Mathematics Examination94.7
MMLU-Pro84.8
Humanity's Last Exam (HLE)22.3
LiveBench65.5
τ²-bench (Tau²-bench)93
IFBench68.1
Terminal-Bench (Hard)31.1
LiveBench - Language66.5
LiveBench - Coding67.4
LiveBench - Math81.1
LiveBench - Instruction Following62
LiveBench - Data Analysis52.3
LiveBench - Reasoning63.5
GPQA Diamond83.8
SciCode42.4
FrontierMath21.4
CorpFin v260.6
TaxEval v271.7