llmboard.ai
Benchmarks
CompareRankings
llmboard.ai
Benchmarks
CompareRankings
HomeModelsKimi K2

Moonshot AI model product

Kimi K2

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.

Updated Aug 10, 2026. Default version: Kimi K2-Instruct-0905

Compare
LLMBoard score51.4Kimi K2-Instruct-0905
Coverage80%12 benchmark families
Context windowN/ATokens
Official input priceN/AOfficial price unavailable

On this page

  • Specification
  • Capability
  • Benchmarks
  • Arena
  • Pricing
  • Versions
  • About
  • Compare
  • Similar models
  • FAQ

Model specification

Structured fields from the published default version.

Version
Kimi K2-Instruct-0905
Released
Sep 5, 2025
Knowledge cutoff
Unknown
Parameters
1T
Context window
N/A
Max output
N/A
Inputs
text
Outputs
text
Open weights
No
License
MIT

Capability profile

This profile uses the latest version under this unique model that has a calculated LLMBoard score. Arena and price are excluded.

Kimi K2-Instruct-0905 category scores

Benchmark results

Published benchmark records for the scored version Kimi K2-Instruct-0905.

29 rows
Columns

Show columns

ACEBench0.8220.0%CAug 7, 2026
AutoLogi0.9220.0%CAug 7, 2026
CNMO 20240.72350.0%CAug 7, 2026
PolyMath-en0.7220.0%CAug 7, 2026
MultiPL-E0.951366.7%CAug 7, 2026
ZebraLogic0.96828.6%CAug 7, 2026
MATH-5001.073280.7%CAug 7, 2026
OJBench0.3990.0%CAug 7, 2026
LiveBench0.8103875.7%CAug 7, 2026
Aider-Polyglot0.6132242.9%CAug 7, 2026
MMLU0.91410086.9%CAug 7, 2026
Multi-Challenge0.5152950.0%CAug 7, 2026
IFEval0.9176575.0%CAug 7, 2026
MMLU-Redux0.9174866.0%CAug 7, 2026
Tau2 Airline0.6172327.3%CAug 7, 2026
Tau2 Retail0.7212620.0%CAug 7, 2026
SuperGPQA0.6223436.4%CAug 7, 2026
Terminal-Bench0.323258.3%CAug 7, 2026
SimpleQA0.3254646.7%CAug 7, 2026
Tau2 Telecom0.7283520.6%CAug 7, 2026
HMMT 20250.430339.4%CAug 7, 2026
SWE-bench Multilingual0.531349.1%CAug 7, 2026
LiveCodeBench0.5407345.8%CAug 7, 2026
AIME 20240.7425321.1%CAug 7, 2026
MMLU-Pro0.84612964.8%CAug 7, 2026
SWE-Bench Verified0.77210431.1%CAug 7, 2026
Humanity's Last Exam0.091921.1%CAug 7, 2026
AIME 20250.51031149.7%CAug 7, 2026
GPQA0.810523355.2%CAug 7, 2026

Arena results

Preference and agent-evaluation signals from published Arena datasets.

No published Arena match

The default version has no published Arena rows, or its source alias has not been resolved.

Pricing

Official vendor API PAYG pricing is summarized first. The table then lists individual provider offerings without treating their minimum as the official price.

Official API
N/A
Official provider
N/A
Lowest third-party
From $1 input, $3 output per 1M via Hugging Face
Tracked offerings
1
1 rows
Columns

Show columns

Hugging Facemoonshotai/Kimi-K2-Instruct-0905global$1$3262.1KAug 7, 2026

Official prices use only the vendor's configured official Provider and positive standard USD PAYG rates. Third-party offers remain explicitly labeled.

Kimi K2 versions

All published versions linked to this unique model. The score columns identify the version used by the current overall ranking.

4 rows
Columns

Show columns

Kimi K2 0905Sep 5, 2025N/A1T262.1K262.1KNoProprietary
Kimi K2-Instruct-0905Sep 5, 202551.41TN/AN/ANoMIT
Kimi K2 BaseJul 11, 2025N/A1TN/AN/ANoMIT
Kimi K2 InstructJul 11, 2025N/A1T200K200KNoMIT

What is Kimi K2?

A concise description based on the published model registry.

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.

Use the benchmark, Arena and pricing sections above as separate evidence. A missing field means the current data snapshot does not support that claim.

Data snapshot: 2026-08-07. Editorial model content is not available in the backend.

Kimi K2 vs nearby models

Open a comparison with the three ranked models immediately above and below this model.

Kimi K2vsLlama 3.1 70BKimi K2vsQwen3 VL 32B ThinkingKimi K2vsQwen2.5 VL 32BKimi K2vsGPT-4-TurboKimi K2vsMercury 2Kimi K2vsPhi 4 Reasoning Plus

Models similar to Kimi K2

Recommendations prioritize the same model type and family, then the closest published LLMBoard score.

#53+8.0
MA

Kimi K2 Thinking

Moonshot AI

59.4 LLMBoard

DetailsCompare
#29+20.2
MA

Kimi K2.5

Moonshot AI

71.6 LLMBoard

DetailsCompare
#7+31.3
MA

Kimi K2.6

Moonshot AI

82.8 LLMBoard

DetailsCompare
#79+0.1
AC

Qwen2.5 VL 32B

Alibaba Cloud / Qwen Team

51.5 LLMBoard

DetailsCompare
#78+0.3
AC

Qwen3 VL 32B Thinking

Alibaba Cloud / Qwen Team

51.7 LLMBoard

DetailsCompare
#81-0.3
OP

GPT-4-Turbo

OpenAI

51.1 LLMBoard

DetailsCompare

FAQ

Common questions about Kimi K2.

When was Kimi K2 released?

Kimi K2's default version was released on Sep 5, 2025.

How much does Kimi K2 cost?

No official standard PAYG price is currently available for Kimi K2. The lowest tracked third-party offer starts at $1 input and $3 output via Hugging Face.

Who created Kimi K2?

Kimi K2 is published under Moonshot AI in the model registry.

What is the context window for Kimi K2?

The current registry does not publish a context window for the default version.

Is Kimi K2 open weight?

No. The default version is not marked as having publicly available weights.

How many API providers offer Kimi K2?

1 published provider offerings are linked to the default version.

What models should I compare Kimi K2 with?

Nearby ranked alternatives include Llama 3.1 70B, Qwen3 VL 32B Thinking, Qwen2.5 VL 32B.

Rankings

OverallCodingText ArenaPricing

Modalities

Image GenerationVideo GenerationSpeech-to-TextEmbeddings

Benchmarks

All BenchmarksReasoningMathCoding

Vendors

All VendorsOpenAIAnthropicGoogle
llmboard.aiCopyright 2026 llmboard.ai