ReelSDK vs fluent-ffmpeg: $2 vs $42 for 1,000 jobs a month

A fluent-ffmpeg alternative after its 2025 deprecation: a 720p transcode and thumbnail for about a fifth of a cent

ReelSDK takes the ffmpeg arguments your fluent-ffmpeg chain builds and runs them over HTTP: turning a 10-second 1080p clip into 720p plus a thumbnail takes about 10 seconds and costs about a fifth of a cent. fluent-ffmpeg was archived on 22 May 2025 and npm flags every version as deprecated, yet it still gets 3.6 million downloads a week; it runs simple jobs on ffmpeg 8.1, but needs ffmpeg and ffprobe on every server, and its README warns it no longer works properly with recent ffmpeg versions.

A month of 1,000 transcodes with thumbnails
  • ReelSDK ffmpeg, billed by the second$2
  • fluent-ffmpeg on a 2 vCPU droplet$42
  • fluent-ffmpeg on an EC2 c7i.large$65

ReelSDK from a measured run: about 6 s of compute per job at $0.02 a minute. The servers are always on at list price, checked 7 November 2026. On a server you already run, fluent-ffmpeg adds no bill, just under 3 hours of 2 CPUs.

Which one to pick

Pick ReelSDK if

  • You'd rather not depend on an archived package, or install and patch ffmpeg and ffprobe on every server and in every Docker image
  • Encodes are slowing down your Node.js app: on our 2 CPUs this small job took about 10 seconds of full CPU
  • You want the same ffmpeg arguments on ffmpeg 8.1 with download URLs, signed webhooks and failed jobs free, at $0.02 per compute minute

Pick fluent-ffmpeg if

  • It works for your jobs today, you pin your ffmpeg version and you're fine owning any fixes
  • You pipe Node streams straight into ffmpeg and out again, without storing files
  • Video must stay on your servers, or you run far more than about 40,000 jobs a month on machines you already pay for

Price per minute

fluent-ffmpeg and ReelSDK, what the same job costs

fluent-ffmpeg itself is free under MIT; the cost is the machine that runs ffmpeg. ReelSDK bills raw ffmpeg at $0.02 per compute minute, by the second, with failed jobs free; the pricing page has every job type. The job: a 10-second 1920x1080 clip transcoded to 720p H.264 at crf 23, plus a 720p thumbnail at 2 seconds.

fluent-ffmpeg and ReelSDK, what the same job costs
OptionWhat it costsPer job1,000 jobs a month
fluent-ffmpeg on your app serverFree; ffmpeg shares the CPU with your app10 s of 2 CPUs$0 extra, under 3 hours of 2 CPUs
fluent-ffmpeg on a DigitalOcean 2 vCPU droplet$0.0625 an hour, $42 a monthUnder a tenth of a cent when busy$42
fluent-ffmpeg on AWS EC2 c7i.large$0.08925 an hourUnder a tenth of a cent when busy$65
ReelSDK ffmpeg$0.02 per compute minuteAbout a fifth of a centAbout $2

Per-job server figures use the 10 s fluent-ffmpeg took on 2 CPUs of our shared test machine. Below about 40,000 of these jobs a month, ReelSDK costs less than keeping a 2 vCPU droplet for them. Prices from fluent-ffmpeg on GitHub, checked November 2026.

Head to head

ReelSDK and fluent-ffmpeg, feature by feature

ReelSDK and fluent-ffmpeg compared
ReelSDKfluent-ffmpeg
StatusMaintained, ffmpeg 8.1Archived 22 May 2025; every npm version deprecated
InstallNothing; HTTPS from any languagenpm i fluent-ffmpeg plus ffmpeg and ffprobe binaries on each server
This jobAbout 10 s from request to both files, about a fifth of a centAbout 10 s on 2 CPUs of our test machine, ffmpeg 8.1
How you write itThe ffmpeg arguments as a JSON arrayChained methods that build ffmpeg arguments
Where it runsFresh machines per job, not your app serverYour server's CPU and disk
Limits500 MB per input, 1 GB per job, 10 minutes per jobYour hardware
ResultsSigned URLs kept 7 days, or a webhook signed with HMAC-SHA256, with up to 6 attempts over about 7 hoursFiles on local disk, or a Node stream
Concurrency3 running jobs and 500 a day per key by defaultAs many ffmpeg processes as your CPUs can take
LicensePay per useMIT; ffmpeg's own license applies to the binary you ship

Be honest

Where fluent-ffmpeg is the better choice

  • Free and already installed

    If fluent-ffmpeg works for your jobs today, it costs nothing to keep, and our test transcode and thumbnail still ran with ffmpeg 8.1 in November 2026. Pin the ffmpeg version and it will likely keep working for simple jobs.

  • Node streams

    fluent-ffmpeg takes a readable stream as input and can pipe output to a stream, so you can transcode an upload as it arrives. ReelSDK works with files: a URL or an upload in, download links out.

  • No limits and no upload

    On your own servers there's no 10-minute job limit, no 500 MB input cap and no upload step, and video never leaves your network.

Migration

Moving a render from fluent-ffmpeg

Each fluent-ffmpeg method ends up as ffmpeg arguments; send those as args to POST /v1/ffmpeg, with the input as a URL. With the request below, the 720p mp4 and its thumbnail came back in about 10 seconds and cost about a fifth of a cent. See the video and the thumbnail. If you'd rather keep ffmpeg local, spawning it with child_process and the same arguments drops the deprecated wrapper.

fluent-ffmpeg today

fluent-ffmpeg.mjs
import ffmpeg from "fluent-ffmpeg";

// Needs ffmpeg and ffprobe on every server, or setFfmpegPath() and setFfprobePath()
ffmpeg("in.mp4")
  .videoFilters("scale=-2:720")
  .videoCodec("libx264")
  .outputOptions(["-crf 23"])
  .audioCodec("aac")
  .on("error", (error) => console.error(error.message))
  .on("end", () => {
    ffmpeg("in.mp4").screenshots({ timestamps: [2], filename: "thumb.jpg", folder: ".", size: "?x720" });
  })
  .save("out.mp4");

The same render on ReelSDK

reelsdk.mjsTry it
const res = await fetch("https://api.reelsdk.com/v1/ffmpeg?wait=60", {
  method: "POST",
  headers: { Authorization: `Bearer ${process.env.REELSDK_KEY}`, "Content-Type": "application/json" },
  body: JSON.stringify({
    "inputs": {
      "in.mp4": "https://api.reelsdk.com/f/lib/demo/landscape.mp4"
    },
    "args": [
      "-i", "in.mp4", "-vf", "scale=-2:720", "-c:v", "libx264", "-crf", "23",
      "-c:a", "aac", "out.mp4", "-ss", "2", "-vf", "scale=-2:720",
      "-frames:v", "1", "thumb.jpg"
    ],
    "outputs": ["out.mp4", "thumb.jpg"]
  }),
});
const { job } = await res.json();

// 200 OK after 10.1 s, trimmed
// {
//   "job": {
//     "id": "job_h25x6de4004l9cu7zys1",
//     "status": "succeeded",
//     "duration_s": 9.782,
//     "outputs": {
//       "out.mp4": "https://api.reelsdk.com/f/out/job_h25x6de4004l9cu7zys1/out.mp4?s=...",
//       "thumb.jpg": "https://api.reelsdk.com/f/out/job_h25x6de4004l9cu7zys1/thumb.jpg?s=..."
//     }
//   }
// }

Field by field

fluent-ffmpeg fields and their ReelSDK equivalents
fluent-ffmpegReelSDK
ffmpeg("in.mp4")inputs: { "in.mp4": url }
.videoFilters("scale=-2:720")"-vf", "scale=-2:720" in args
.videoCodec("libx264").outputOptions(["-crf 23"])"-c:v", "libx264", "-crf", "23"
.screenshots({ timestamps: [2] })A second output: "-ss", "2", "-frames:v", "1", "thumb.jpg"
.save("out.mp4")outputs: ["out.mp4", "thumb.jpg"] as download URLs
.on("end") and .on("error")job.status, error.code, or a webhook

FAQ

ReelSDK vs fluent-ffmpeg, answered

Is fluent-ffmpeg deprecated?

Yes. Its README says it is deprecated, no longer maintained and no longer works properly with recent ffmpeg versions, and the repository was archived on 22 May 2025. npm marks every version, up to the latest 2.1.3, as deprecated. It still gets about 3.6 million downloads a week.

Does fluent-ffmpeg still work with new ffmpeg versions?

For simple jobs, often yes: our transcode, thumbnail and ffprobe call all ran with ffmpeg 8.1 and fluent-ffmpeg 2.1.3, but nobody fixes bugs or adds options anymore, so anything that breaks stays broken unless you patch it yourself.

What should I use instead of fluent-ffmpeg in Node.js?

To keep ffmpeg local, call it with child_process.spawn and pass the arguments as an array; it's free and has no wrapper to go stale. To stop running ffmpeg yourself, ReelSDK takes the same arguments over HTTP, runs them on ffmpeg 8.1 and returns download URLs.

How much does ReelSDK charge for an ffmpeg job?

$0.02 per compute minute, billed by the second, and failed jobs are free. This 720p transcode with a thumbnail used 6.3 seconds of compute, about a fifth of a cent, so 1,000 of them cost about $2.

Other comparisons

Ready to build?

Ship video from your app today

Your first $0.50 is free, and one key covers every endpoint.

Moonlit lake with a lit cabin under the mountains