Перейти к основному содержанию
AITier

Kimi K2

KimiKimiОткрытые весаMIT · Коммерческое использование

Описание

Kimi K2-Instruct-0905 is the latest, most capable version of Kimi K2, achieving state-of-the-art performance in frontier knowledge, math, and coding among non-thinking models. This Mixture-of-Experts model features 32 billion activated parameters and 1 trillion total parameters, meticulously optimized for agentic tasks. Key features include enhanced agentic coding intelligence, extended context length to 256K tokens, and a hybrid architecture trained with MuonClip optimizer on 15.5T tokens. The model achieves 65.8% on SWE-bench Verified (single attempt), 47.3% on SWE-bench Multilingual, and excels at tool use with 70.6% on Tau2-retail. It is a reflex-grade model without long thinking, designed to act and execute complex tasks seamlessly.

Дата выхода
2025-07-11
Параметры
1.0T
Длина контекста
131K
Модальности
text

Радар способностей

37
general
51
coding
68
reasoning
48
science
60
agents
0
multimodal

Рейтинги

Домен#МестоОценкаИсточник
Агентные возможности102
39.0
LS
Рейтинг кодинга231
50.0
AA
Общий рейтинг240
51.0
AA
Наука255
49.0
AA

Оценки бенчмарков (LLM Stats)

(LLM Stats (zeroeval))

Agents

Terminal-Bench30.0%Сам.
Terminus25.0%Сам.

Biology

GPQANYU + Cohere + Anthropic (2023)75.1%Сам.

Chemistry

SuperGPQA57.2%Сам.

Code

HumanEvalOpenAI (2021)93.3%Сам.
EvalPlus0.80 / 100Сам.
SWE-bench Verified (Agentic Coding)65.8%Сам.
SWE-Bench Verified65.8%Сам.
Aider-Polyglot60.0%Сам.
LiveCodeBench53.7%Сам.
SWE-bench Multilingual47.3%Сам.

Communication

Tau2 Retail70.6%Сам.
Tau2 Telecom65.8%Сам.
Tau2 Airline56.5%Сам.
Multi-Challenge54.1%Сам.

Factuality

SimpleQA31.0%Сам.

Finance

MMLU89.5%Сам.
MMLU-Pro81.1%Сам.
ACEBench76.5%Сам.

General

MMLU-Redux92.7%Сам.
C-Eval92.5%Сам.
MMLU-redux-2.090.2%Сам.
IFEvalGoogle Research (2023)89.8%Сам.
MultiPL-E85.7%Сам.
TriviaQA85.1%Сам.
CSimpleQA78.4%Сам.
LiveBench76.4%Сам.
LiveCodeBench v653.7%Сам.
SWE-bench Verified (Agentless)51.8%Сам.

Math

MATH-50097.4%Сам.
GSM8k97.3%Сам.
CBNSL95.6%Сам.
CNMO 202474.3%Сам.
MATH70.2%Сам.
AIME 202469.6%Сам.
PolyMath-en65.1%Сам.
AIME 202549.5%Сам.
HMMT 202538.8%Сам.
Humanity's Last Exam4.7%Сам.

Reasoning

AutoLogi89.5%Сам.
ZebraLogic89.0%Сам.
HumanEval-ER81.1%Сам.
MuSR76.4%Сам.
SWE-bench Verified (Multiple Attempts)71.6%Сам.
OJBench27.1%Сам.

Индексы оценки AA

(Artificial Analysis)
Math Index(Artificial Analysis)
57.0
Intelligence Index(Artificial Analysis)
19.7
Math 500(OpenAI (2024), subset of Hendrycks et al. MATH (2021))
1.0
Mmlu Pro(TIGER-Lab (Univ. of Waterloo, Toronto, CMU, 2024))
0.8
Gpqa(NYU + Cohere + Anthropic (2023))
0.8
Aime(MAA (Mathematical Association of America))
0.7
Tau2(Sierra + U Toronto + Vector Institute (2025))
0.6
Aime 25(MAA (Mathematical Association of America))
0.6
Livecodebench(UC Berkeley + MIT + Cornell (2024))
0.6
Lcr(Artificial Analysis)
0.5
Ifbench(Google Research (2023))
0.4
Scicode(UIUC + Argonne National Lab (2024))
0.3
Terminalbench Hard(Stanford × Laude Institute (2026))
0.2
Hle(Center for AI Safety + Scale AI (2025))
0.1

Оценки категорий LLM Stats

(LLM Stats (zeroeval))
Structured Output
90
Instruction Following
90
Language
90
Legal
80
Finance
80
Healthcare
80
Biology
80
Math
70
Physics
70
Frontend Development
70
Chemistry
70
Reasoning
60
General
60
Communication
60
Economics
60
Tool Calling
60
Code
50
Factuality
30
Agents
20
Vision
0

Цены

Цена ввода$0.57 / 1M токенов
Цена вывода$2.3 / 1M токенов
Смешанная цена (3:1)$1.002 / 1M токенов

Скорость

Токенов/сек0.0
Задержка первого токена0.00s
Время до первого ответа0.00s

Рейтинг цен провайдеров

Рейтинг цен провайдеров

7 провайдеров

Самый дешевый: OpenCode ZenСамый дорогой: LLM Gateway
ПровайдерВводВывод
1OpenCode ZenСамый дешевый
$0.4
$2.5
2FastRouter
$0.55
$2.2
3KimiОсновной
$0.57
$2.3
4OpenRouter
$0.57
$2.3
5Kilo Gateway
$0.57
$2.3
6Vercel AI Gateway
$0.57
$2.3
7LLM Gateway
$0.57
$2.3

Сравнение цен разных API-провайдеров для этой модели.

Внешние ссылки

LLM StatsArtificial Analysis