Scan to join WeChat groupKimi K2.5
Kimi K2.5 for coding, long-running tasks and agent workflows
Kimi K2.5 is a Moonshot/Kimi model for coding, long-running tasks and agent workflows, often evaluated for Chinese document and knowledge workflows.
starsCapabilities
visibilityVision understandingcodeFunction callingstreamStreaming outputdata_objectStructured output
paymentsContext and pricing
Context limit262,144
Max output262,144
Knowledge cutoff2025-01
Input price$0.6/ 1M tokens
Output price$3/ 1M tokens
Cached input price$0.1/ 1M tokens
descriptionOverview
Overview
Kimi K2.5 is listed in Moonshot AI's official platform documentation, with model ID kimi-k2.5. Kimi models are commonly used for Chinese long-text reading, document Q&A, content generation and knowledge-base assistants.
Best for
Use Kimi K2.5 for document uploads, report analysis, knowledge-base Q&A, Chinese long-form text and office assistants. Test context length, document understanding, Chinese summarization quality, cost and latency before production.
lightbulbUse cases
- Chinese long-document analysis
- Knowledge Q&A and summarization
- Office assistants and content generation
- Retrieval-augmented reading
thumb_upStrengths
- Good fit for Chinese long-text tasks
- Clear Kimi product positioning
- Easy to tier by context length
- Useful for document and knowledge products
infoLimitations
- Cost and latency vary by context tier
- Reasoning and coding need separate evaluation
- Document quality depends on input format
- Limits depend on Moonshot documentation