Skip to main content

Kimi K3 (max)

KimiKimiOpen WeightKimi K3 License · Commercial OK

Description

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.

Release Date
2026-07-16
Parameters
2.8T
Context Length
1.0M
Modalities
image, text, video

Capability Radar

57
general
74
coding
93
reasoning
72
science
70
agents
80
multimodal

Rankings

Domain#RankScoreSource
Agentic Capability9
63.0
LS
Code Ranking2
98.0
AA
General Ranking6
95.0
AA
Multimodal Ranking7
70.0
LS
Science8
93.0
AA

Benchmark Scores (LLM Stats)

(LLM Stats (zeroeval))

Agents

AA-Briefcase1548.00 / 3000
Program Bench77.8%SR
DECK-Bench73.5%SR
Toolathlon73.2%SR
Kimi Code Bench v272.9%SR
OfficeQA Pro63.3%SR
Job Bench52.9%SR
MLS-Bench Lite48.3%SR
PostTrainBench36.6%SR
SpreadsheetBench 234.8%SR
AutomationBench30.8%SR

Code

FrontierSWE81.2%SR
DeepSWE 1.169.0%
DeepSWE67.5%SR
SWE-Marathon42.0%SR

Math

MathVision97.8%SR

Reasoning

GPQANYU + Cohere + Anthropic (2023)93.5%SR
CharXiv-R91.3%SR
BrowseCompOpenAI (2025)91.2%SR
Terminal-Bench 2.188.3%SR
MCP Atlas84.2%SR
Humanity's Last Exam56.0%SR
APEX-Agents37.6%SR

Search

DeepSearchQA95.0%SR

Vision

OmniDocBench91.1%SR
BabyVision85.7%SR
MMMU-Pro (with tools)83.4%SR
MMMU-Pro81.6%SR
PerceptionBench58.5%SR
WorldVQA51.0%SR
ZEROBench0.41 / 100SR

AA Evaluation Indices

(Artificial Analysis)
Gpqa(NYU + Cohere + Anthropic (2023))
93.5
Terminalbench V2 1
85.0
Lcr(Artificial Analysis)
82.7
Coding Index(Artificial Analysis)
76.2
Intelligence Index(Artificial Analysis)
59.7
Scicode(UIUC + Argonne National Lab (2024))
58.7
Hle(Center for AI Safety + Scale AI (2025))
46.9
Tau Banking
46.0

LLM Stats Category Scores

(LLM Stats (zeroeval))
Productivity
100
Agents
92
Reasoning
78
General
60
Physics
90
Search
90
Biology
90
Chemistry
90
Math
80
Multimodal
80
Code
70
Tool Calling
70
Vision
70
Systems
40

Pricing

Input Price$3 / 1M tokens
Output Price$15 / 1M tokens
Blended Price (3:1)$6 / 1M tokens
Cache Read Price$0.3 / 1M tokens

Speed

Tokens/sec36.4
Time to First Token3.56s
Time to Answer58.53s

Provider Price Ranking

Provider Price Ranking

25 providers

Cheapest: FireworksMost Expensive: Tinfoil
ProviderInputOutput
1FireworksCheapest
$0
$0.00002
2Moonshot AI
$0
$0.00002
3Together
$0
$0.00002
4Novita
$0
$0.00002
5CrofAI
$2
$8
6Requesty
$2.25
$11.25
7Vancine
$2.4
$12
8DevPass (LLM Gateway)
$2.83
$14.13
9DigitalOcean
$2.85
$14.25
10KimiPRIMARY
$3
$15
11OpenCode Go
$3
$15
12Vivgrid
$3
$15
13GitHub Copilot
$3
$15
14OpenCode Zen
$3
$15
15AIHubMix
$3
$15
16Moonshot AI (China)
$3
$15
17Cortecs
$3
$14.999
18Neuralwatt
$3
$15
19Neon
$3
$15
20EmpirioLabs AI
$3
$15
21CoralBricks
$3
$15
22Charm Hyper
$3.2664
$16.332
23Venice AI
$3.75
$18.75
24GreenPT
$3.762
$18.81
25Tinfoil
$4
$20

Compare pricing across different API providers for this model.

External Sources