Kimi K3 (Max)
KimiKimi开源权重Kimi K3 License · 商用许可
描述
Kimi K3 is Moonshot AI's flagship open Mixture-of-Experts model for long-horizon coding, knowledge work, and reasoning. It has 2.8 trillion total parameters, activates 16 of 896 experts, and combines Kimi Delta Attention, Attention Residuals, and Stable LatentMoE. The model supports a 1M-token context window, native image and video understanding, tool calling, structured output, and always-on reasoning with max thinking effort at launch.
发布日期
2026-07-16
参数规模
2.8T
上下文长度
1.0M
支持模态
image, text, video
能力雷达图
44
general
74
coding
93
reasoning
72
science
70
agents
80
multimodal
排行榜排名
基准测试分数 (LLM Stats)
(LLM Stats (zeroeval))Agents
AA-Briefcase
1548.00 / 3000
Program Bench
77.8%自报
DECK-Bench
73.5%自报
Toolathlon
73.2%自报
Kimi Code Bench v2
72.9%自报
OfficeQA Pro
63.3%自报
Job Bench
52.9%自报
MLS-Bench Lite
48.3%自报
PostTrainBench
36.6%自报
SpreadsheetBench 2
34.8%自报
AutomationBench
30.8%自报
Code
FrontierSWE
81.2%自报
DeepSWE 1.1
69.0%
DeepSWE
67.5%自报
SWE-Marathon
42.0%自报
Math
MathVision
97.8%自报
Reasoning
GPQANYU + Cohere + Anthropic (2023)
93.5%自报
CharXiv-R
91.3%自报
BrowseCompOpenAI (2025)
91.2%自报
Terminal-Bench 2.1
88.3%自报
MCP Atlas
84.2%自报
Humanity's Last Exam
56.0%自报
APEX-Agents
37.6%自报
Search
DeepSearchQA
95.0%自报
Vision
OmniDocBench
91.1%自报
BabyVision
85.7%自报
MMMU-Pro (with tools)
83.4%自报
MMMU-Pro
81.6%自报
PerceptionBench
58.5%自报
WorldVQA
51.0%自报
ZEROBench
0.41 / 100自报
AA 评测指数
(Artificial Analysis)Gpqa(NYU + Cohere + Anthropic (2023))93.5
Lcr(Artificial Analysis)88.7
Terminalbench V2 185.0
Coding Index(Artificial Analysis)76.2
Scicode(UIUC + Argonne National Lab (2024))59.5
Hle(Center for AI Safety + Scale AI (2025))46.9
Tau Banking46.0
Intelligence Index(Artificial Analysis)43.6
Terminalbench V4 012.6
LLM Stats 分类评分
(LLM Stats (zeroeval))Productivity100
Agents92
Reasoning78
General60
Physics90
Search90
Biology90
Chemistry90
Math80
Multimodal80
Code70
Tool Calling70
Vision70
Systems40
定价
输入价格$3 / 1M tokens
输出价格$15 / 1M tokens
混合价格(3:1)$6 / 1M tokens
缓存读取价格$0.3 / 1M tokens
缓存写入价格$3 / 1M tokens
速度
Tokens/秒44.7
首Token延迟3.44s
首回答延迟48.22s
供应商价格排行
供应商价格排行
6 个供应商
最便宜: DeepInfra最贵: Kimi
供应商输入输出
1DeepInfra最便宜
$0
$0.00001
2Fireworks
$0
$0.00002
3Together
$0
$0.00002
4Moonshot AI
$0
$0.00002
5Novita
$0
$0.00002
6Kimi主要
$3
$15
比较该模型在不同 API 供应商之间的定价。