跳轉到主要內容

Kimi K2.5 (Non-reasoning)

KimiKimi開源權重MIT · 商用許可

描述

Kimi K2.5 is Moonshot AI's flagship agentic model and a new SOTA open model. It unifies vision and text, thinking and non-thinking modes, and single-agent and multi-agent execution into one model. Built with Full-Parameter RL tuning, it achieves state-of-the-art performance across agents, coding, image, and video benchmarks.

發布日期
2026-01-27
參數規模
1.0T
上下文長度
262K
支援模態
image, text, video

能力雷達圖

26
general
40
coding
79
reasoning
52
science
50
agents
80
multimodal

排行榜排名

領域#排名分數來源
智慧體能力模型榜130
33.0
LS
程式碼能力榜205
54.0
AA
通用能力榜208
55.0
AA
多模態榜27
57.0
LS
科學能力194
56.0
AA

基準測試分數 (LLM Stats)

(LLM Stats (zeroeval))

Agents

WideSearch79.0%自報
DeepSearchQA77.1%自報
BrowseCompOpenAI (2025)74.9%自報
PaperBench63.5%自報
Terminal-Bench 2.0Stanford × Laude Institute (2026)50.8%自報
SWE-Bench ProPrinceton NLP (2024)50.7%自報
CyberGym41.3%自報
FrontierSWE26.0%

Biology

GPQANYU + Cohere + Anthropic (2023)87.6%自報
SciCode48.7%自報

Code

SWE-Bench Verified76.8%自報
SWE-bench Multilingual73.0%自報
OJBench (C++)57.4%自報

Economics

FinSearchComp T2&T367.8%自報

Finance

MMLU-Pro87.1%自報

General

LiveCodeBench v685.0%自報
MMMU-Pro78.5%自報
SimpleVQA0.71 / 100自報
LiveBench69.1%
LongBench v261.0%自報

Healthcare

VideoMMMU86.6%自報

Image To Text

OCRBench92.3%自報

Long Context

LongVideoBench79.8%自報
LVBench75.9%自報
AA-LCR70.0%自報

Math

AIME 202596.1%自報
HMMT 202595.4%自報
MathVista-Mini90.1%自報
MathVision84.2%自報
IMO-AnswerBench81.8%自報
Humanity's Last Exam50.2%自報

Multimodal

InfoVQAtest92.6%自報
OmniDocBench 1.588.8%自報
Video-MME87.4%自報
MMVU80.4%自報
CharXiv-R77.5%自報
MotionBench70.4%自報
WorldVQA46.3%自報
ZEROBench0.11 / 100自報

Reasoning

Seal-057.4%自報

AA 評測指數

(Artificial Analysis)
Intelligence Index(Artificial Analysis)
30.1
Tau2(Sierra + U Toronto + Vector Institute (2025))
0.8
Gpqa(NYU + Cohere + Anthropic (2023))
0.8
Lcr(Artificial Analysis)
0.7
Ifbench(Google Research (2023))
0.4
Scicode(UIUC + Argonne National Lab (2024))
0.4
Terminalbench Hard(Stanford × Laude Institute (2026))
0.2
Hle(Center for AI Safety + Scale AI (2025))
0.1

LLM Stats 分類評分

(LLM Stats (zeroeval))
Legal
90
Language
90
Finance
90
Long Context
80
Math
80
Image To Text
80
Frontend Development
80
Video
80
Multimodal
70
Physics
70
Reasoning
70
Search
70
Structured Output
70
General
70
Healthcare
70
Biology
70
Chemistry
70
Vision
70
Agents
60
Code
50
Tool Calling
50
Safety
40

定價

輸入價格$0.6 / 1M tokens
輸出價格$3 / 1M tokens
混合價格(3:1)$1.2 / 1M tokens
快取讀取價格$0.095 / 1M tokens

速度

Tokens/秒0.0
首Token延遲0.00s
首回答延遲0.00s

供應商價格排行

供應商價格排行

11 個供應商

最便宜: NanoGPT最貴: Ofox
供應商輸入輸出
1NanoGPT最便宜
$0.3
$1.9
2TensorX
$0.5
$2.8
3OpenRouter
$0.57
$2.85
4ZenMux
$0.58
$3.02
5Kimi主要
$0.6
$3
6Jiekou.AI
$0.6
$3
7NovitaAI
$0.6
$3
8Kilo Gateway
$0.6
$3
9Vercel AI Gateway
$0.6
$3
10HPC-AI
$0.6
$3
11Ofox
$0.6
$3

比較該模型在不同 API 供應商之間的定價。

外部連結