Machine translation is metered by the number of source characters sent to a translation engine. Plans that include machine translation come with a character quota, and this page is where you see how much of it you have used.

Find it on your organization, under Billing → API Usage.

Your Current Period#

At the top of the page, a progress bar shows the characters you have used against your quota for the current billing period, with the percentage used. It turns amber as you pass 90% of your quota, and red once you go over it.

The billing period runs from the start of your subscription, or from the point at which overuse was last invoiced. Organizations without a subscription — a free trial, for instance — are measured over a rolling 30-day window instead.

Your quota comes from your plan. If we have agreed a specific limit with you, that limit applies instead of the plan’s.

History#

Below the quota bar, a table breaks your usage down by month and by engine, covering the last twelve months. Each engine — Google Translate, Microsoft Translator, Gemini and Mistral — is listed separately, so you can see which one is actually consuming your quota.

Click any month to drill into a daily breakdown for that month. This is the quickest way to explain a sudden jump in usage: a day-level spike almost always corresponds to a bulk Batch Operation or a large file upload followed by machine translation.

Going Over Your Quota#

Exceeding your quota does not switch machine translation off. Overuse is allowed and is simply billed.

• When you first pass your quota, the account owner receives an e-mail.
• The page shows how many characters you are over, and the resulting charge.
• Overuse is charged at €30 per million characters.
• The charge is added as a line item on your next invoice, described as “Machine Translation overuse”, after which the overuse counter resets for the new period.

If your usage falls back below the quota before the period ends, the alert clears itself and there is nothing to pay.

Keeping Usage Down#

If you are consuming more than you expected:

• Check the per-engine breakdown. The AI engines are the ones typically used for the whole-file and plural-aware translation described in Machine Translation, and they are correspondingly easy to run up a bill with.
• Look at what Batch Operations are doing. Machine-translating an entire language in one operation is a large, single, very visible consumption of quota.
• Consider Auto-Translate, which reuses translations you already own across your organization’s projects. It costs no machine translation characters at all, and for a company with several products sharing terminology it will often cover a large share of new segments.
• Remember that the Translation Memory suggestions are free too.

A Note on the Name#

This page is labelled “API Usage” in the interface for historical reasons, but what it measures is machine translation characters. It is not a record of your calls to the WebTranslateIt API — API rate limits are documented separately in the API overview.