Kimi tools

Kimikimi_reading

Kimi reading assistant

Read text, webpages and documents, extract insights and pay by actual token usage.

Start with your agent

Claude CodeCodexCursor

Download and install the dedicated Mouter Skill for “Kimi reading assistant”: https://mouter.ai/skills/kimi_reading.zip

Pricing

How credits work
RateBilling unitCredits
Uncached inputPer 1000 tokens0.003
Output tokensPer 1000 tokens0.015
Cached inputPer 1000 tokens0.0003

Initial hold (credits): 0.001. The final charge is based on tokens used.

Usage example

Agent

Example input

Use kimi_reading. Read this report and summarize its theme, key points and conclusions: <document or web URL> Confirm inputs and estimated cost before running.

Example output

Example only: Kimi reading assistant has not been run. Once a task succeeds, its result contains these fields:

  1. texttext
  2. finish_reasontext
  3. usageobject: prompt_tokens, completion_tokens, cached_tokens, total_tokens

Values depend on your input. The matching JSON example structure is in the API tab.

  • Getting started: supply text or up to five public webpage/document URLs. State the questions and desired response structure, then set an output budget if needed.

FAQ

Can Kimi Reading read text, webpages and documents?

Provide text of up to 100,000 characters, or up to 5 public webpage or document URLs in document_urls. Use prompt to describe what to extract or summarize, with 1–8,000 characters. Replace example URLs with accessible sources before running.

How do I control the answer length?

max_tokens accepts 256–8,192 and defaults to 4,096. The result contains text, finish_reason and usage. Check finish_reason when deciding whether the response finished normally or reached its length limit.

Which token usage is returned for billing?

usage includes prompt_tokens, completion_tokens, cached_tokens and total_tokens. Billing uses the published input, cached-input and output rates with actual token usage. The initial hold is separate from the final charge.

How is Kimi reading assistant billed?

Published reference rates: Uncached input: 0.003 credits · Per 1000 tokens; Output tokens: 0.015 credits · Per 1000 tokens; Cached input: 0.0003 credits · Per 1000 tokens. Initial hold (credits): 0.001. The final charge is based on tokens used. Final charges use your account’s rates and actual usage.