Kimi K2.6 has been released with improved long-context coding stability

Build with Kimi API
Transform Energy into Intelligence

Access state-of-the-art large language models designed for complex software automation, deep research, and intelligent agents at unprecedented speed and scale.

Get Started → View Models

Trusted by millions of professional developers

Vercel TRAE Kilo Code OpenRouter Windsurf Flowith Cursor Coze CodeBuddy
Latest Models

Built for tasks of any scale

Delivering fast, reliable performance with high quality results across every workload.

kimi-k2.6

K2.6

Kimi's latest and most intelligent model, possessing stronger and more stable long-term code writing capabilities and significantly enhanced Agent autonomous execution capabilities.

Cache Hit: $0.16 / MTok Input: $0.95 / MTok Output: $4.00 / MTok
kimi-k2.5

K2.5

Supports visual and text input, thinking and non-thinking modes, and dialogue and agent tasks across a wide range of use cases.

Cache Hit: $0.10 / MTok Input: $0.60 / MTok Output: $3.00 / MTok
kimi-k2

K2 0905

MoE architecture base model with exceptional code and Agent capabilities. 1T total parameters with 32B active parameters for optimal performance.

Cache Hit: $0.15 / MTok Input: $0.60 / MTok Output: $2.50 / MTok
Official Tools

Plug and Play

Reliable, production ready tools — integrate once and start building.

Web Search

Access latest information and cite authoritative sources.

Rethink

Intelligent idea organization tool.

Memory

Persistent conversation history and user preference storage.

Excel

Analysis tool for Excel and CSV files.

Code-Runner

Python code execution environment.

Quick JS

Safe JavaScript execution via QuickJS engine.

Fetch

URL content extraction and markdown formatting.

Convert

Unit conversion supporting physics and currency.

Base64

Encoding and decoding utility.

Date

Date and time processing tool.

Random-Choice

Random selection utility.

Fast Execution for Complex Workflows

A complete tool ecosystem

Integrating general intelligence with vertical models — built for real world complexity.

Agent Programming

Built for complex software automation. Based on Kimi K2's instruction following and logical reasoning capabilities, Kimi enables autonomous programming agents to handle debugging, refactoring, and multi-step development workflows with high precision.

Deep Research & Reasoning

Designed for long horizon, complex reasoning tasks. Leveraging the K2 Thinking model's reasoning capability and up to 300-step tool calling, Kimi acts as an expert-level research assistant supporting strategic research, pricing and competitive analysis, stock and crypto investing analysis.

Legal Intelligence

Streamline high-stakes document workflows. Kimi is optimized for rigorous attention to detail, ideal for legal and IP scenarios — from automated contract review to complex patent analysis and drafting.

Conversation Intelligence

Extracting deep value from unstructured conversation data. Kimi excels in high fidelity conversation analysis, supporting use cases from quality assessment and public opinion monitoring to customer intent detection.

Powering Innovation Through Partnership

Breaking resource boundaries

Unlocking exponential growth in core value — aligning to succeed together.

#AgenticCoding
KiloClaw

Kimi offers SOTA-level performance at a fraction of the cost. It's tremendously good at long-context tasks across the codebase, as well as the day-to-day work needed to support an always-on agent.

#AIAgents
Genspark

Runs its agent platform on Kimi K2 (0905), turning crisp instructions into autonomous agents that solve real world problems.

#AgenticCoding
CodeBuddy

Integrates Kimi K2 Thinking as its underlying model, providing core capabilities to help developers solve complex programming tasks and improve R&D efficiency.

#AIforScience
XtalPi

Integrates with the Kimi K2.5 model, instantly reading chemical formulas and charts, extracting key points from cutting-edge literature, and accelerating drug discovery.

#FinancialResearch
AlphaEngine

Integrates with Kimi K2 Thinking to create a FinGPT Agent. Supports over 300 tool calls, deeply analyzes massive amounts of research reports and financial data.

#AIforScience
DP Technology

Highlights Kimi K2.5's dominance in chemical literature understanding — ranking Top 2 on RxnBench benchmarks alongside closed-source leaders like Gemini-3 and GPT-5.2.

Quick Start

Launch your first AI workflow in just a few steps

Flexible pricing and enterprise solutions to match your scale.

Pay-As-You-Go

Ideal for individual developers and small teams
  • Flexible pay-as-you-go billing with transparent and controllable costs
  • Lightning fast access — one click Kimi model calls with no setup required
  • Tiered usage benefits — automatically unlock higher rate limits based on cumulative spend

Get Started

Enterprise Solutions

Designed for medium and large enterprises
  • High performance architecture with flexible rate limits and multi-project support
  • Security and compliance assurance with SLA-backed reliability
  • Dedicated technical support with priority response and model optimization

Contact Sales