FlowData

Why is credit reserved, and what makes up the cost of a media job?

Published on 10. November 2026

Alongside your Credit, the FlowData billing page shows Reserved and Available. Reservations come from media jobs. This article explains why part of your credit is reserved and how the cost of a job is made up.

How a media job is charged

Each output of a job is charged separately:

  • Video and audio by started minutes of the finished output, with a minimum of one minute. An output of 61 seconds counts as two minutes, one of 59 seconds as one.
  • Images, thumbnails and posters per output.

For video, the price per minute depends on the resolution of the output, more precisely on its shorter side (Video SD, Video HD, Video 4K). The tiers are listed in How are FlowData prices calculated?.

If a job produces several outputs, each one counts on its own. For example, a job with the ugc_video pipeline turns a video of 2:10 minutes into one version with a watermark, one without, and a thumbnail. Both versions are 540 pixels on the shorter side. That is 3 SD video minutes for the first version, 3 for the second, and 1 image.

Only outputs that were successfully delivered are charged. If an output fails, it costs nothing. If a job stops before any output is finished, it costs nothing either. Status requests are free.

The current prices per minute and per image are on the pricing page, and the prices that apply to your team are under Billing.

Why is credit reserved?

When you start a job, FlowData does not yet know how long your video is. To stop a job from failing halfway through for lack of credit, FlowData reserves an amount up front that is certain to cover it. You can see this reservation under Billing, at Reserved. Only the amount under Available can be used for new jobs.

A reservation works like this:

  1. At the start, FlowData reserves the longest duration the pipeline allows for each video and audio output. That is 10 minutes for regular video pipelines and 3 hours for long video and audio. Images are reserved at one unit per output. If the size of a video output is still open, the highest tier is used as a precaution.
  2. After analysing the file, FlowData knows the real duration. If the video is longer than assumed, the reservation is increased. If your credit or limit does not cover that, the job stops with insufficient_quota or spend_limit_exceeded, and nothing is charged.
  3. At the end, only the actual amount for the delivered outputs is posted, based on their real duration and resolution. The rest of the reservation is released immediately.

This is why a lot may be reserved at first for a short video. In the end you only pay what the outputs actually cost.

How to keep reservations small

Send the file size in bytes when you start the job, for example "input": {"url": "https://example.com/video.mp4", "size": 48211904}. FlowData then estimates the duration from the size, with a small safety margin, instead of reserving the maximum. With a direct upload, FlowData takes over the declared size automatically.

This is especially worthwhile if

  • you are on prepaid and your credit is running low,
  • many jobs run at the same time,
  • a key has its own Monthly limit. Jobs in progress count against that limit with their reservation, so a key would otherwise hit its limit sooner than necessary.

Good to know

  • A job is charged at the prices in force when it was created. A later price change does not affect jobs in progress.
  • FlowData credit is separate from FlowAI API credit. You cannot use one for the other.
  • On postpaid, the billing page shows Reserved for jobs in progress. There, the reservation counts towards your credit line and monthly limit.
  • How to top up credit and set a spend limit is explained in How does billing work in FlowData?.

Was this article helpful?

Back to Billing & invoices