API Billing Guide

Kimi API is billed by token usage. Prices vary by model and feature.

Billing Basics

• Billed by token: Each API call is billed separately for input token and output token usage.
• token unit: 1M = 1,000,000 tokens.
• Prices vary by model: More capable models have higher unit prices. Choose the right model based on your business needs.

Add-on Feature Billing

• Online search: Each call to the online search feature incurs an additional ¥0.03 charge, independent of token usage.

Context Caching

Context Caching lets you cache frequently used context, such as system prompts and reference documents. token usage for cache hits is billed at a discounted rate, helping reduce the cost of repeated context.

For detailed Context Caching pricing, please refer to the official documentation.

Billing Details

For the complete model price list and billing rules, visit:

https://platform.kimi.com/docs/pricing/chat

Cost Optimization Tips

• Set the max_tokens parameter appropriately to avoid generating long, unnecessary content.
• Use Context Caching to cache repeated system prompts and context.
• Choose a model based on task complexity. Lightweight models are enough for simple tasks.
• Keep prompt design concise to reduce unnecessary input token usage.