Lorka AI logoLorka AI logo

Kimi Models List (2026): Every Moonshot AI Model and What It’s Best For

Moonshot AI has developed several Kimi models for coding, reasoning, knowledge work, visual understanding, and agents. Kimi K3 is its flagship model, while Kimi K2.7 Code specializes in development, and Kimi K2.6 offers flexible multimodal capabilities. Compare each model’s strengths, context window, inputs, reasoning features, and pricing.

Try Kimi Models on Lorka
Lorka AI icon
Ask anything...

Kimi Models at a Glance

The current Kimi models range from frontier reasoning to fast coding and general multimodal work. Use the model cards below to compare their capabilities, inputs, context windows, reasoning modes, and Kimi API pricing.

Kimi K3 Family

⚡

Kimi K3

🤖Best for long-horizon coding, deep reasoning, and end-to-end knowledge work

July 2026

Kimi K3 is Moonshot AI’s flagship model for its most demanding coding, reasoning, and knowledge-based tasks. Its 1M-token context window enables it to handle large codebases, lengthy technical documents, and extensive research materials in a single session.

Inputs: Text, Images
Outputs: Text
Context: Up to 1M tokens
Price (input): (cached input): $0.30, (uncached input): $3.00 per 1M tokens
Price (output): $15.00 per 1M tokens
Learn more about Kimi K3 →

Kimi Coding Models

💭

Kimi K2.7 Code

🧠 Best Kimi model for reliable long-context coding and instruction following

June 2026

Kimi K2.7 Code is designed for complex, multi-turn development workflows, including repository changes, debugging, refactoring, issue resolution, and software agents.

Inputs: Text, images, and video
Outputs: Text
Context: Up to 256K tokens
Price (input): (cached input): $0.19, (uncached input): $0.95 per 1M tokens
Price (output): $4.00 per 1M tokens
⚡

Kimi K2.7 Code HighSpeed

⚡Best Kimi model for fast coding agents and real-time developer workflows

June 2026

Kimi K2.7 Code HighSpeed provides the same coding capabilities as the standard model, with faster output for pair programming, rapid debugging, and iteration-heavy agent workflows.

Inputs: Text, images, and video
Outputs: Text
Context: Up to 256K tokens
Price (input): (cached input): $0.38, (uncached input): $1.90 per 1M tokens
Price (output): $8.00 per 1M tokens
💡

Kimi K2.6

💡Best Kimi model for multimodal reasoning and flexible agent workflows

April 2026

Kimi K2.6 is a general-purpose model for document analysis, research, visual tasks, video understanding, dialogue, and AI agents, with optional thinking for greater speed.

Inputs: Text, Images
Outputs: Text
Context: Up to 256K tokens
Price (input): (cached input): $0.16, (uncached input): $0.95 per 1M tokens
Price (output): $4.00 per 1M tokens
Learn more about Kimi K2.6 →

Legacy Kimi Models

Moonshot V1 is Moonshot AI’s classic-generation model line, previously offered with 8K, 32K, and 128K context windows. Kimi K2.5 is also an earlier multimodal model that has been replaced by newer Kimi releases.

How to Choose the Right Kimi Model

Use this Kimi model comparison to figure out the best tool for your needs, whether it's coding, writing, math, or other types of tasks.

  • If you need Moonshot AI’s most capable model for complex knowledge work: Choose Kimi K3.
  • If you need a 1M-token context window: Choose Kimi K3.
  • If you need reliable coding with persistent reasoning: Choose Kimi K2.7 Code.
  • If you need faster output for interactive coding agents: Choose Kimi K2.7 Code HighSpeed.
  • If you need general multimodal reasoning and flexible agent workflows: Choose Kimi K2.6.
  • If you need a model for large repositories and long development tasks: Choose Kimi K3 for maximum capability or Kimi K2.7 Code for a specialized coding experience.

Compare Kimi with Claude models, GPT, Gemini, DeepSeek, and Llama in a single workspace

Try Kimi Now

Kimi Model FAQs

Kimi K3 is the latest flagship model from Moonshot AI. It is designed for long-horizon coding and end-to-end knowledge work, with native visual understanding and a context window of up to 1 million tokens.