Uploads and files

Per upload

100 MB

Per key

20 GB a day

resets at midnight UTC

Kept for

7 days

then the URL returns 404

Upload a file#

Send the raw file as the request body, with its name in the filename query parameter.

curl -sS "https://api.reelsdk.com/v1/uploads?filename=river.jpg" \
  -H "Authorization: Bearer $REELSDK_KEY" \
  --data-binary @river.jpg

The answer is 201 Created with the file's URL, its size, the content type we picked and when it expires.

{
  "url": "https://api.reelsdk.com/f/in/fr78av0etm46huw3/river.jpg",
  "bytes": 71069,
  "content_type": "image/jpeg",
  "expires_at": "2026-11-02T01:15:29.712Z"
}

Pass url as any input: video_url for captions, an entry in images for a slideshow, a voiceover or music track, or a file in ffmpeg inputs.

Parameters#

We keep the last part of the name, swap anything other than letters, digits, ., - and _ for _, and cut it to 64 characters. My Trip (1).MOV becomes My_Trip_1_.MOV.

File types#

The extension decides the type. Anything else returns 400 UNSUPPORTED_FILE_TYPE.

File URLs#

Uploads live under /f/in/ and job outputs under /f/out/. Both work without an API key, so you can put them in a video tag, hand them to a CDN or share them with your users.

• GET and HEAD are supported, with Access-Control-Allow-Origin: * so browsers can fetch them.
• Range requests answer 206 Partial Content, so players can seek and resume. A range past the end answers 416.
• Every file has an ETag. Send it in If-None-Match and an unchanged file answers 304.
• Images, video and audio are served inline, so they play in the browser. Subtitles, JSON and text download as attachments.
• Cache-Control: public, max-age=604800, immutable, since a file never changes once written.

curl -sS -r 0-99 -D - -o /dev/null \
  "https://api.reelsdk.com/f/in/fr78av0etm46huw3/river.jpg"

# HTTP/2 206
# content-type: image/jpeg
# content-range: bytes 0-99/71069
# etag: "cc07182fe39b32813949820bce71df61"
# cache-control: public, max-age=604800, immutable

Retention#

Uploads and outputs are deleted 7 days after they're written, and their URLs then return 404 with file not found or expired. The upload response tells you when in expires_at. Copy outputs you want to keep to your own storage, for example from your webhook handler.

Remote URLs#

Every input also takes any public http or https URL, including signed links from S3, R2, Google Cloud Storage or Supabase, as long as they stay valid until the job has downloaded them.

• URLs can be up to 4,096 characters. localhost, private and reserved IP addresses and hostnames without a dot are refused with 400.
• We follow up to 5 redirects and try each download up to 3 times when the server answers 5xx, 408 or 429 or the connection drops.
• A download that sends no data for 30 seconds counts as a failed try.
• Each file can be up to 500 MB, and all inputs of a job together up to 1,000 MB.
• Outputs of one job work as inputs to the next, so you can caption a slideshow without downloading it in between.

A URL that keeps failing ends the job with 422 INPUT_FETCH_FAILED and a message such as Could not fetch video_url: HTTP 404.

Errors#

Last updated 26 October 2026