Repository navigation
Costs and Limits
Preen has no fee of its own. You use your own API key. Your provider charges you for what Preen sends. This page gives no prices, because providers change them. Look at your provider's price page. Need a key? See Getting an API key.
- Preen sends text to the model. The model reads it in small pieces called tokens.
- Preen counts the tokens for every message.
- Preen has no price built in. You can type your provider's Price per million input tokens. Then Preen shows an estimate in dollars.
- Test connection fills in the price for you when your provider reports a cost.
- Without a price, Preen shows tokens only.
Once Preen has a price, Settings shows a line like "At your price, sorting costs about $X per 100 emails and $Y per 1,000 emails." Preen works it out from your price. It also uses the average tokens your past messages used. These numbers are estimates. Your provider's bill is the real cost.
If you have not sorted any mail yet, Preen has no average. It counts only what your categories add. The line then says "at least about". It gets better as you use Preen.
The Estimated monthly cost limit box stays locked until there is a price. A note beside it says so. The note says Test connection can fill in the price. Later, a Use $N button can offer a starting limit. It is the estimate for 1,000 emails, rounded up. The button only fills the box. Nothing is saved until you click Save.
Select several messages in the toolbar pop-up. It shows the expected tokens and cost before you click Classify selected or Preview selected. The estimate comes from your past use. Try it on the last 7 days shows the number of messages and the estimate in a dialog first. You can say Not now.
You can see your use in two places:
- The toolbar pop-up shows today's messages and tokens, and this month's tokens.
- The activity page shows this month's totals, a chart by category and a chart of tokens by month.
Deep sends part of the message text. That is more tokens than Surface. The settings help says Deep was clearly more accurate in testing, especially for junk and real people, for only a little more cost. See Settings.
| Limit | What it does | Default |
|---|---|---|
| Messages per day | Stops sending when the day's count is reached | 500 |
| Estimated monthly cost limit | Stops running when this month's estimate reaches it | none |
| Speed | Sets how fast Preen works and how many messages one automatic run takes | Medium |
| Your provider's credit limit | The hard stop. It is set at your provider, not in Preen | up to you |
Every request counts toward the limits. That includes automatic runs, right-click, folders and the toolbar button.
Preen checks the limits before each message. A request already running may finish after you reach a limit.
When a limit is reached, Preen says so. For example: "Daily limit of 500 messages reached (change it in Settings → Limits & allowance)". The toolbar icon also shows a mark. Messages left over stay waiting for a later run. The activity page tells you how many were not processed.
The day and month follow your computer's local time.
If you select more than 100 messages to classify or preview, Preen asks you to confirm. The window shows how many messages and a token estimate, based on your average so far. Your daily limit still applies.
| Speed | How it works | Messages in one automatic run |
|---|---|---|
| Low | One message at a time, with a 2-second rest after each | 10 |
| Medium | One message at a time, with a short rest after each | 25 |
| High | Two at a time, no rest | 50 |
| No limit | Three at a time, no rest | 100 |
Medium or Low is best for large folders. Runs take longer, but your computer stays quick. "No limit" is not recommended for large folders or slow computers.
If the provider is busy, slow or cannot be reached, Preen tries again. It waits 2 seconds, then 4 seconds. A request that gets no answer in 20 seconds is counted as a time-out. A wrong key is not retried.
A message that keeps failing is tried in up to 3 automatic runs. Then Preen gives up on it, so it does not spend tokens again and again.
- Preview a few messages while you learn. Previews still use tokens.
- Use Review this run to apply a preview. It reuses the answers Preen already has, so it sends nothing new and costs nothing more. See Getting started.
- Preen skips mail you corrected by hand, so it does not spend tokens on it again. You can still choose Classify anyway.
- Start with "Inbox only" for each account.
- Lower Messages per day.
- Set an Estimated monthly cost limit, and also a credit limit on your key.
- Use Surface for accounts where you do not need the text read.
- Use Only mail from the last days to keep Preen from reading old mail.