Kimi K2
Moonshot AI · 2025
Moonshot AI's latest model with 128K context, strong search integration, and reasoning abilities.
Quick Facts
Parameters
Undisclosed
Context Window
128K tokens
Modalities
text, image
Open Source
No
Pricing
Free tier / API from ~$0.30/1M tokens
Released
2025
Developer
Moonshot AI
About
Kimi K2 is Moonshot AI's (月之暗面) latest large language model, building on the success of the Kimi series with enhanced reasoning, search integration, and long-context processing. It features a 128K token context window and deep integration with live web search, enabling real-time information retrieval and synthesis. Kimi K2 excels at processing Chinese content, research tasks, and complex queries requiring up-to-date information. Available through the Kimi chat web interface and API, it has gained significant popularity among Chinese users for its strong performance and user-friendly design.
Strengths
- +Strong Chinese language processing and understanding
- +Real-time web search integration for current information
- +128K context for long document analysis
- +User-friendly interface with file upload support
Weaknesses
- −Limited availability outside China
- −Smaller global user base
- −English performance below top Western models
Best For
Chinese-language research and analysis
Real-time information retrieval and synthesis
Long document processing with search
China-market AI assistant applications
Pricing
Free (Web)
$0
- Kimi K2 chat
- Web search
- File uploads
API
From ~$0.30/1M tokens
- Pay-as-you-go
- 128K context
- Search integration
Technical Specs
Parameters
Undisclosed
Context Window
128K tokens
Modalities
text, image
Languages
Open Source
No
Developer
Moonshot AI
Released: 2025
Related Models
GPT-4o
OpenAI
OpenAI's flagship multimodal model combining text, vision, and audio in one unified interface.
GPT-5
OpenAI
OpenAI's latest flagship model with enhanced reasoning, larger context, and improved multimodality.
Claude 3.5 Sonnet
Anthropic
Anthropic's balanced model offering strong reasoning, coding, and long-context capabilities.
Claude 4 Opus
Anthropic
Anthropic's most powerful model for complex reasoning, research, and specialized tasks.