0

Kimi K2.5

Kimi K2.5 is an AI model from Kimi.

Publisher
Moonshot AI
Family
Kimi
Context
262K
Input
Output
License
Modified MIT
Released
27 Jan 2026

Cite

Notes

Only stored in your browser.

Attribution

Attribution policy →
Context
262K
Weights
Open
downloadable
Input
$0.60
/1M tok
Output
$2.75
/1M tok
Blended
$1.14
#397/ 499/1M · 3:1
Speed
0
#103/ 498tok/s
Cache read
$0.15
0.25x input

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

Reasoning effortIndexPriceSpeed
Reasoning23.5$1.140 t/s
Non-reasoning19.4$1.200 t/s

Intelligence Index per reasoning-effort setting, via Artificial Analysis

Reported on 37 evals across 13 domains - best 95.9% on τ²-bench (Tau²-bench) · #14 of 321

Reported eval scores

37
τ²-bench (Tau²-bench)95.9
OSWorld-Verified63.3
CorpFin v268.3
TaxEval v274.2
MortgageTax66.5
Finance Agent (v2)35.8
LiveBench - Coding77.9
LiveBench - Language77.7
Terminal-Bench (Hard)34.8
IFBench70.2
LiveBench - Math84.9
LiveBench - Reasoning76
SWE-bench Multilingual67.3
LiveBench - Data Analysis61.4
LiveBench - Instruction Following57.4
GPQA Diamond87.9
SciCode49
LiveBench72.5
Humanity's Last Exam (HLE)30.7
Vals Index26.3

Head-to-heads

29

benchmark comparisons against other models

vs Claude Opus 4.631 shared evalsvs Gemini 3.1 Pro Preview30 shared evalsvs GPT-5.229 shared evalsvs Qwen3.6 Plus29 shared evalsvs Claude Sonnet 4.528 shared evalsvs GPT-5.528 shared evalsvs Claude Opus 4.828 shared evalsvs GPT-5.428 shared evalsvs Claude Opus 4.728 shared evalsvs Claude Sonnet 4.627 shared evalsvs Claude Opus 4.526 shared evalsvs Claude Fable 526 shared evalsvs GLM 5.126 shared evalsvs GPT-5 Mini25 shared evalsvs DeepSeek V3.225 shared evalsvs GPT-5.4 Mini25 shared evalsvs Grok 4.1 Fast24 shared evalsvs Claude 4 Sonnet23 shared evalsvs GPT-5 Nano22 shared evalsvs Gemini 2.5 Pro21 shared evalsvs GLM 4.720 shared evalsvs gpt-oss-120b19 shared evalsvs GPT-518 shared evalsvs GPT-4o15 shared evalsvs GPT-6 Astra13 shared evalsvs GPT-4.1 Mini12 shared evalsvs GPT-4.112 shared evalsvs GPT-4o-mini12 shared evalsvs GPT-4.1 Nano12 shared evals