[Docs] Clarify docs about the usage and rate limit

What feature would you like to see?

According to the docs, kimi code plan:

每 5 小时支持约 300–1200 次请求 (~300-1200 requests per 5 hrs)

The phrase "300-1200 requests" may be ambiguous, leading readers to mistakenly believe that the number of requests corresponds strictly and solely to the tier of the plan they purchased. However, this data is calculated based on the number of tokens consumed (and is rather rough even exaggerated).

If all your work is to ask AI agent to write you a bubble sort, or ask it to introduce what they are capable of, then yes, you can probably work all day long without ever reaching the limit. But with a little bit of harness - AGENT, MEMORY, REFERENCE and SPEC - applied, you will soon used up all your auotas since the real world engineering actually CONTAINS context.

I hope this part of the docs could be modified to better address the issue/concern; or better, add a section to clearly list available tokens for each tier.

Additional information

No response

The issue does not name a documentation file or test. Locate the Kimi Code plan section describing requests per five hours, then review how usage and token limits are documented elsewhere. Done means the wording makes token-based variability clear and accurately explains or lists tier limits if that information is available.