Plans & Pricing
Usage limits
Every account has three independent usage limits. Each request must remain within all three limits, which are measured using rolling windows rather than fixed calendar periods.
The three ceilings
| Limit | Window | What counts | Purpose |
|---|---|---|---|
| Messages per 5 hours | Rolling 5 hours | Successful model responses | Limits bursts within a session. |
| Messages per week | Rolling 7 days | Successful model responses | Manages sustained usage across sessions. |
| Requests per minute | Rolling 60 seconds | API requests | Prevents excessive request bursts. |
The five-hour and weekly limits are calculated whenever a request is made. Capacity returns gradually as older messages leave each window. It does not reset all at once on the hour, day, or week.
How the limits work together
For Free, Pro, Max, and Team plans, the five-hour allowance is one tenth of the weekly allowance. Enterprise provides additional weekly capacity to support sustained team workloads.
| Plan | Messages / 5h | Messages / week | Ratio |
|---|---|---|---|
| Free | 20 | 200 | 1:10 |
| Pro | 100 | 1,000 | 1:10 |
| Max | 1,000 | 10,000 | 1:10 |
| Team | 300 per seat | 3,000 per seat | 1:10 |
| Enterprise | 1,200 per seat | 20,000 per seat | 1:16.7 |
These limits allow steady usage throughout the week while controlling short periods of unusually high demand. Enterprise includes proportionally more weekly capacity for sustained workloads across larger teams.
Pooled limits
Team and Enterprise limits are pooled across the account rather than assigned to individual members. Each seat adds capacity to a shared pool that any member can use.
For example, a five-seat Team account shares 1,500 messages per five-hour window and 15,000 messages per week. Pooling prevents unused capacity from being locked to individual members, although one member can use a larger share of the total. The console provides account wide usage information to help teams monitor this.
Viewing your usage
The Usage in the console shows your remaining messages for both rolling windows and when capacity will become available again. These values use the same usage data that enforces your limits, so they remain in sync.
When a limit is reached
If you reach a limit, the request is rejected and the client identifies which limit has been reached. Autohand Code also provides a direct link to the upgrade page in the console. Requests are not queued or automatically switched to a smaller model.
Because the five-hour and weekly limits are independent, you may still have weekly messages available after reaching the five-hour limit. This is the most common reason a session pauses during a task.