Question about Kimi Code subscription quota scaling between Allegretto (¥199) and Allegro (¥699)
Hello Kimi team,
I have a question about how the Kimi Code quota scaling works between the Allegretto (¥199/month) and Allegro (¥699/month) plans.
According to the subscription page:
-
Allegretto (¥199/month): Kimi Code 20× quota
-
Allegro (¥699/month): Kimi Code 60× quota
So the Allegro plan appears to provide 3× the Kimi Code quota compared with Allegretto.
However, based on actual usage data, I am not seeing a 3× difference in the 5-hour usage window.
Example 1: Allegretto (¥199/month) user Image 2
A user reported that they only consumed about half of one 5-hour window, with the following token usage:
-
Input Tokens: 37.15M
-
Cached Tokens: 36.34M
-
Output Tokens: 63.9K
Using Kimi K3 API pricing as a reference, this is approximately:
~$14.3 worth of API usage
If the user had fully consumed the entire 5-hour window, the estimated usage would be:
~$28–30 API value
Example 2: My Allegro (¥699/month) usage Image 3
My usage after a full 5-hour window (actually slightly exceeded):
-
Input Tokens: 8.40M
-
Cached Tokens: 37.78M
-
Output Tokens: 114K
-
Total Tokens: 46.3M
Estimated API value:
~$38 API usage
Comparison
| Plan | Estimated full 5-hour window usage |
|---|---|
| Allegretto ¥199 | ~ $28–30 |
| Allegro ¥699 | ~ $38(actually slightly exceeded) |
My questions:
-
Does the 20× / 60× Kimi Code quota multiplier refer to:
-
5-hour window capacity?
-
daily/monthly cumulative quota?
-
number of requests?
-
another internal quota system?
-
-
Is the Allegro plan actually expected to allow approximately 3× the usage within the same 5-hour window, or does the 3× multiplier only apply to total accumulated usage over time?
-
Could there be differences in quota consumption depending on:
-
cache hit rate?
-
model selection (Kimi K3 vs other models)?
-
coding agent behavior?
-
output token generation?
-
I am trying to understand whether my Allegro plan is working as expected, or if I am misunderstanding how the quota multiplier is calculated.
Thank you!
